ThE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

345 E. 47th St., New York, N.Y. 10017 98-GT-347

S re Society shall not be responsible for statements or opinions advanced in papers or dtsausskon at meetings of the Society at #3i oftn[
® sections, or printed in its publications. Discussion is printed only if the paper is published in an ASME Journal. Authorization to.phatocct
for internal or personal use is granted to libraries and other users registered with the Copyright Clearance Center (CCC) provided
$3/article or $4/page is paid to CCC, 222 Rosewood Dr., Danvers, MA 01823. Requests for special permission or bulk reproduction
should be addressed to the ASME Technical Publishing Department.

Copyright o 1998 by ASME All Rights Reserved Printed in USA.

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

GAS TURBINE PARAMETER CORRECTIONS

Allan J. Volponi
Senior Principal Engineer
Hamilton Standard, a Division of United Technologies Corp.

ABSTRACT m Mass (Ibm)

The various parameters appearing along an engine's gas path, such as N Speed (rpm)
flows, pressures, temperatures, speeds, etc., vary not only with power P Pressure (psia)
condition but also with the ambient conditions at the engine's inlet. P Arbitrary gas path parameter
Since a change in inlet temperature and/or pressure will contribute to P Arbitrary gas path parameter corrected to standard day
an attendant change in a gas path parameter's value, it would be conditions
difficult to characterize the aero-thermodynamic relationships between Q Torque
gas turbine engine parameters, (even at a constant engine operating R Gas constant
point) unless the ambient conditions are somehow accounted for. This S Entropy
is usually accomplished through the use of corrected engine T Temperature (degrees R)
parameters. Although most of these corrections are well known by Wa Air mass flow (lbs/sec)
practitioners in the industry, knowledge of their origin does not Wf Fuel mass flow (lbs/sec)
appear to be as commonplace. The purpose of this paper is to fill that 0 nondimensional temperature (ratio)
gap and furnish a summary of the commonly used corrections for the S nondimensional pressure (ratio)
"major" gas path parameters that are used in performance analysis, p Density
diagnostics and control design, and to offer a derivation of these K Fluid thermal conductivity
corrections. We will suggest both an analytic approach as well as an v Fluid velocity
empirical approach. The latter can be used to establish the correction µ Fluid viscosity
for parameters not directly addressed in this paper, as well as to fine rl Adiabatic efficiency
tune the correction factors when actual engine data is available. IF Flow capacity

NOMENCLATURE Subscripts
A Area 1,2,3,... Engine station number
cp Specific heat @ constant pressure (BTU/lbm °R) s Static quantity
cv Specific heat @ constant volume (BTU/lbm °R) t Total quantity
D Diameter
Fn Net thrust (lbs) Superscripts
g Acceleration constant ( 32 ft/sec 2 ) * Corrected quantity
H Heat transfer coefficient (BTU/hr ft 2 °R)
HP Horsepower HISTORICAL PERSPECTIVE
h Enthalpy The use of corrected parameters to describe the performance of a gas
I Moment of inertia turbine appears to be as old as the machine itself [Warner (1945),
J Mechanical equivalent of heat (778 ft-lbs/BTU) Sanders (1946), Shepherd (1949)]. Earlier works [e.g. Capon, (1930)]
k Ratio of specific heats ( cp Icy) reference many of the same parameters as applied to more general

Presented at the International Gas Turbine & Aeroengine Congress & Exhibition
Stockholm, Sweden — June 2-June 5, 1998
This paper has been accepted for publication in the Transactions of the ASME
Discussion of it will be accepted at ASME Headquarters until September 30,1998
aero-thermodynamic machines such as air compressors. These and dependent variable
typically take the form of so-called dimensionless quantities derivable
from dimensional analysis, which has been the primary technique for Fn Thrust ML/t2
establishing what I will refer to as the classical parameter corrections.
Since this technique is somewhat old and is not necessarily known to where t=Time, T=temperature (degrees), M=Mass, and L=Length.
all, the analytic derivations that are provided in this paper will use a The total number of fundamental dimensions in this case is 4.
more general method which relies only on a knowledge of basic
calculus and a handful of thermodynamic principles. Since Thus if we assume that we have a relationship of the form Fn = f(N,
dimensional analysis has played such an important historical role in T2, F2, v, D, R), which implies an implicit form f(Fn , N, T2, P2
establishing these corrections, we will describe (briefly) its principle v, D, R) = 0, involving 7 quantities. By virtue of the it-theorem, we

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

and derive a particular correction as an illustrative example. The may express the relationship in terms of 7-4=3 dimensionless
following description follows that given by [Buckingham (1914)]. quantities as follows:

Let us assume that we have a physical relationship to be represented

q ( .1 ,r 2+ ir 3) =0
by n physical quantities Ql, Q2, •..,Qn. Without any loss of
where
generality, we may write this relationship as
=NP2 T ' D ` R d ,;r2=vP2 Tf DgR " ,2r3=F,Pk T2 D " R '
.f1(Q,,Q2,...>Qn) = 0 (1)
The zti terms are dimensionless and hence their units must be equal to
Since physical quantities are measured in physical units (feet, seconds, LO MO TO t0. For example
lbs, degrees, etc.) or some algebraic combination of units, it can be
shown that the above general relationship must have the form 1 2

^' ( )t( Lt - l a TbLc (t2 T)

° =L M°T°t°
MQQb2...Q,^ =0 = Lc+2d-a Ma Tb-d t -1-2a-2d

where M is a dimensionless number and each of the products being 0=a

summed must have the same units (otherwise you could not add them
together). If we divide both sides by any one term in the sum, the 0=c+2d—a
expression takes the form O=b—d
0 =1 + 2a + 2d which implies a = 0
^ NQi' Qi2 ...1^" + 1= 0 b=d=-1/2
c=-1
where each product in the sum is a dimensionless quantity. If we
represent each of these terms by zti, then we may rewrite this equation This yields a parameter of the form
as
N/(L RT2 )
M7r,, 7 2 , ..., a») = 0 (2)
Repeating this process for the other dimensionless quantities, yields
Loosely stated, the Buckingham it—Theorem declares that the number the following relationship
of dimensionless products lti necessary for Equation (2) to represent
Equation (1) is n — k , where k is the number of fundamental
dimensions used to define the variables.
F
P2 A
_N v
— fi D RT
2 ' RT2 J
Applying the above principles, let us assume that we have a single
spool turbojet engine having the following independent parameters and if we assume a constant geometry (A=const, D=const) we obtain
(and dimensions)

N
Parameter
spool speed (rpm) l/t
Dimension F
[ __ Mn
Plf2(_
^ L r
sF =f3 N, Mn)J
Tl ambient temperature T
P1 ambient pressure M/(Lt2) which would imply that the appropriate corrections for net thrust and
v aircraft velocity L/t
D compressor diameter L spool speed are F/ Sand N/^ respectively.
R gas constant L2/(t2T)

2
 INTRODUCTION such as fuel flow (Wf), for example, are also corrected for humidity
In the sequel, we will assume a two shaft, mixed flow, turbofan and fuel heating value. The impact of changes in viscosity with
engine with station numbers as depicted in Figure 1 below. This altitude (Reynolds effects) or changes in the gas composition and
reference is made for convenience only, and all corrections that are their impact on gas path parameters will not be considered in this
given in this paper can be applied equally to all gas turbine engines. paper. A discussion of these types of corrections are beyond the
scope of this presentation and thus, we will assume the above form
(Eq. 3) for all corrected parameters p*.

The values for a and b will, in general, vary with engine type and
cycle, however, there are some values which might be considered

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

standard classical corrections and are approximations which are
commonly used in practice for all gas turbines.Table I depicts many
of the common gas path parameters and their standard day corrections.
In the sequel we will attempt to develop all of these and show the
classical value as a special case, where applicable.
T3 T9
P53 PS/ PS
FIGURE 1: TWIN SPOOL MIXED FLOW TURBOFAN Table 1: Common Gas Turbine Parameter Corrections
Parameter a b Corrected Symbol
Rotor Speed 0.5 0 N
N* =—
For any gas path parameter P, the equivalent corrected parameter will
be denoted by P* throughout this discussion. In general, a change in Airflow -0.5 1
the engine inlet conditions T2 and P2 will be accompanied by an w
Wa^
attendant change in any downstream gas path parameter . A S
Fuelflow a 1
Wf * = Wf = Wf
corrected parameter P* would be constant regardless of the change in
inlet condition and represents the value the parameter P would have at (classical: 0.5)
some fixed reference inlet condition. This reference condition can be
Thrust 0 1 Fn
whatever you like, however, it is common practice to select standard Fn*=
day conditions (T2 = 518.67 deg R, P2 = 14.696 psia) for this 5
purpose. Thus, we can assume without loss of generality that P _ HorsePower 0.5 1 HP
f( T2 , P2, P*). Thus it follows that HP*=
^S
Torque 0 1
dTz+ *
-(TTz ) (TPT)^z+ laP*) Q* 5
Temperature a 0 T T
(classical:1.0)
d^ _ r BP / P l dTz dPz J 8P / P 'tip * 7 = =5
_ +1_____1 Pressure 0 1
P l 8T2 / Tz J 9•.conal
h-conr Tz l aP2 / Pz ) r2 =—, Pz
=conv P*=P
;D.—,t
S
Acceleration 0 1
1V*=N
Now if we make the simplifying assumption that the first two partials S
are constant (the third clearly is constant and equal to unity) and Metal Temp 0.74 0.8 T
denote them by a and b respectively, then Rate Tm* =
8 o7' S'0s

d^ za TZ+b dPZ+ dP
P T2 P2 P* DERIVATIONS
In the sequel, we shall attempt to supply some rationale for these
If we now define, the dimensionless parameters 0 = T2/518.67 and 6 = corrections. Most will be motivated from simple thermodynamic
P2114.696, then we arrive at relationships and simplifying assumptions and do not require an
extensive knowledge of either thermodynamics or gas turbine
operation. Where applicable we will attempt to supply the required
Ae--a--b— ;a*^ (3) definitions.
P e 6 0a8b

Thus, a corrected parameter maintains the same units and is calculated The most extensively used and perhaps the simplest correction of all
by means of theta and delta exponent corrections. We might mention is the correction for temperatures, and is the first one which we shall
at this juncture, that Equation (3) is not only approximate for the consider.
reasons already alluded to, but in addition, some gas path parameters

K3
 Corrected Temperature: T / 6
0= d.I S= J (dh
T
– R j'—=c (dT –R rdP
J P pJ T J P
As a general rule, it seems reasonable that engine inlet temperature 2 2 2 2 2
(T2 ) should have a direct effect on downstream temperatures in the
engine's gas path. It seems natural, for example, that at a given P31
=cp in► T3 I–Rln(p2
operating condition, increasing T2 would in turn increase T3, which in l T2 ) JJ)
turn would increase T4, etc. To motivate this concept, we need to
consider the T-S diagram. To be definite, let us consider the R/cp
compression induced by the Fan and LPC from station 2 to station 3 T3
of our sample engine in Figure 1. Without loss of generality, an T2 (P2 )

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

isentropic compression has been assumed. Figure 2 depicts a
compression from P2 to P3 starting at different temperatures (T2 and Likewise, integrating between 2* and 3* with the same assumption on
T2*). specific heats, provides

P
T * R/cp
3*
P T2 * \P2)
3
– T
2* T2

S
T3
Figure 2: Compression T-S Diagram T3*= T3
(T2 /T2 *) 0
At this point we will note two properties of the T-S diagram which we
will state without proof. (The reader can derive these for him/herself, as required.
armed only with the elementary laws of thermodynamics.) The
relationships are Corrected Pressure: P / S

1. The curves of constant pressure are monotonically The pressure changes experienced at various stations throughout the
increasing with S engine's gas-path are the effect of either compressions or expansions
2. The constant pressure curves diverge from one another resulting from the action of the engine's turbomachinery. For
with increasing S purposes of motivating the correction for pressure, we can consider
without loss of generality, a compression process; the argument for
These properties suggest that the * quantities are larger in magnitude expansion pressures would be similar.
than their non-starred counterparts and that ( T3 – T2 )< ( T3* – T2*
). What we will now show, however, is that the ratio of T3 over T2 is Let us consider, for example, the pressure P3 at the exit of the LPC.
approximately equal to T3* over T2* ! This pressure is related to temperature and pressure by virtue of the
following relationship
To accomplish this we need one more relationship from basic
thermodynamics (which follows from the definition of entropy and k-1

T2 =
the perfect gas law).
where ri = polytropic efficiency
`2)
dS–dh–RdP
k = c p / ç, = ratio of specific heats
T P

where S, h, T, P, R denote entropy, enthalpy, temperature, pressure and Thus, since T3/9 = constant implies T3/T2 = constant, we have that
the gas constant, respectively. Integrating this relationship between P3/P2 = constant, which in turn implies that P3/6 = constant, which
states 2 and 3 and assuming constant specific heat, yields, establishes the correction at station 3. Implicit in this argument is that
the ratio of specific heats k and efficiency rl do not change. Moving
downstream to station 4 we have that

4
 kn krn for Mn const., thus providing a corrected quantity.t

1 3 ) k1
P3(T (T3 16) k1 — coast.
Corrected Fuel Flow: WF / SJ
Consider the (simplified) energy equation bounding the combustor
PPP
a 1'3 = const
S P3 6
BTU/ 1 /{,
= 'Ib9WF = (Wa+WFJ Lh= (Wa+WF)(h5 — h4)
sec
Thus, we proceed downstream and establish in general that P16 =
constant = P.

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

where h denotes enthalpy, q denotes the heating value of the fuel and
i denotes the adiabatic efficiency of the combustor. If we assume that
Corrected Mass Flow: Wa j/8 (at a given operating point) efficiency and fuel heating value are
constant, then if we take logs of both sides and differentiate, we
Consider flow Wa (in pps) moving with uniform velocity v through a obtain
pipe having cross sectional area A (ft 2 ). Then
dWF = d(Wa +WF ) + d(Lh)
Wa =p A v = density(lbs / ft 3 ) x area(ft 2 ) x velocity (ft / sec) WF (Wa + WF ) Ah

For an ideal gas, we also know that P s = pRT, from which it follows
_ Wa dWa
(_W l dW F F

R. + WF)) Wa l (Wa + W4) WF

that
Wa = p V = sPv +I Cp s ) —(C 4 4
d
A RT,. )

and from the definition of mach number Mn and a little algebraic manipulation gives

Mn = v / gkRT, Wa l dWF Wa 1 dWa ( cps Ts dT5 cpa Ta dTa

Wa+WF/WF `(Wa+WF)JWa + I Oh)TS -( ^h )T4
we obtain the relationship
Now, at a given operating point, we can assume that the following
Wa = Ps Mn gkR7Wa T = r Pf 1XK n relationships hold:
^

A RT AP lP J 'MFR
W
S =coast Wa +1/2 T
d a - 4 =0
We know, however, that for an isentropic process Wa T4 P4

l e
a =
and esconst a =T
d 2 = s
k-1 T4 T2T5
T =1+ kI
2kMn 2 and p =(1+ k-1 Mn 2 I
P4
—= const = a = 2
4 P2
Substituting, we obtain Thus,

l k+1
dWF = dP2 2 + Wa + WF cps T dT2 Wa + WF Cp4 T dT2
I1+ k2 1 M2I
2(1-k) -1 / 2 dT
= RM \
°APrTn n/
WF P2T Wa ^h) T2 - ( Wa^h , T2

and multiplying both sides by ( A/Tt )1/2 / (6/Pt ) to get the proper
units (pps) yields
dp2
L
+ I 1+ yFa
/ lp
c^h 5
- ^a
cp
J -1/2 J(5)
T d 2

dP2 dT
k+1 _ +x
Wa . A gk(T /9) ( k-1 212(1-k) P2T2
Mn 11 2 Mn J =Constant
S P/r S R WF = const
50X

t T/B and P18 are corrected parameters and hence constant by virtue of the
previous derivations.

5
 Taking logs of both sides and differentiating, produces
Furthermore, if we make the assumption that specific heats are
constant, i.e. cp 4 = cp5 then the exponent x reduces to 1/2 + Fuel- dF„ = (W$ / g) 2gJ_, dW$ 1 d6,h
Air Ratio = (approximately) 1/2 and we obtain the classical
F. F. [W$ + 2 AhI
normalization
+ "9l P9 — P') 49
`A9 + I P9 P9 P9 – ( 7_
– Pz I ^9 _ ^z
Pz

WF
Fn yz Pz
J J
Sytl
(Ws / g) 2gJOh dW$l k / dhs l r..2&2)l h^
hs9
dhg

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

F. [0's l 2Ah) h
Corrected Net Thrust: F„ / 6

Expressions for the net thrust produced by a gas turbine engine will
depend upon the configuration of the engine, i.e., whether it is a
turbojet or a turbofan, mixed flow or non-mixed flow, as well as an +
A 9 (P9 – Ps9)

F,,
9
A9
+ 1_p9l dP9
P9–P2/ P9
Pz
.P9 –P2 ) P2
l ^z
zOM
expression for ram drag as a function of mach number. To keep the (fixed geometry)
derivation as simple as possible, we will assume a static thrust
expression, i.e., mach = 0, no ram drag and, we will once again refer
to our sample engine configuration which is that of a mixed flow __ (W8 / g) 2gJAh dW$ cp,Ts dT$ cp9Ts9 dT 1 l
turbofan. It should be noted that derivation that follows can be
repeated (in spirit) with more complicated expressions for net thrust
F. LW$ + 2ih T$ ( 2Nz J Ts9^ J
for other engine types and flight conditions to arrive at similar + `^I (P9 — Ps9 P9 dP9 (TP2 ^2
representations for corrected thrust. F. ) (P9 –P2 J P9 –Pz ) P2 ,
The thrust produced by this engine depends on the momentum of the
air ejected from the exhaust nozzle as well as the net force (due to Now, at a given engine operating point (steady state), the following
conditions hold
pressure) acting across the total area of the nozzle. In symbols

W e$ dW$ 1 dT dP$
W9
g v 9 +A 9 (P9 –PA ) constant + - -0
–
F. — -- -

S$ W 27; F
.
where
T$ = constant dT –z
— =0
A2 T T
W9 = W$ = .Total airflow at nozzle
A 9 = .Nozzle Area 9
T = constant = T
d – dT
2 =0
02 T
P9 = .Nozzle exit total pressure
P59 = P2 = .Nozzle exit static pressure (ambient) Ps =constant ---=0
$ 2
62 P8 P2
v 9 = 2gJ(h9 – hs9 ) = . Velocity at exhaust nozzle
h9 = .Total (stagnation) enthalpy at exit P9 = constant dP9
--- =0
S2 P9 P2
hs9 = .Static enthalpy at exit
J = .Mechanical equivalent of heat Substituting, we obtain
.(778 ft-lbs/BTU)

The expression for the exit velocity follows directly from the dF„ – (W8 / g) 2gJih dP$ cp$T1 cp9 T g _ dTz
definition of stagnation (total) enthalpy. Similarly, there is an F. F„ P8 + ( 2Ah 20h 2)7;
expression for the velocity at station 8 (mixer) involving total and
x
static enthalpies at station 8. The flow from station 8 to 9 (through a
nozzle) we will assume to be adiabatic, i.e., no work done, no heat + .i(P9 – P:9) dPz
added. By conservation of energy, the total entaiphies are equal (h8 = F„ P2
h9). If we denote the change in enthalpy (hg - hsg) = (h8 - hs9) = Ah ,
then we may rewrite the above expression for thrust as = (W$ l ^
g) r^a +x 'zl + ``^i(P9 – P.,^) ^z
F„LP$ T2J F; P2
rt Y2
F„ = w
$ 2gJ^ h + A 9 (P9 – PP )
g
1 dP2d7;
dT dP2 dT + =1
_ (y1 + y2) -+ ylx - _ +y,x- — since Mach = constant
N M„ P2T P2 T 2T 2T
=> F" = constant
dN I d9 = 0
S B YIX 22 N - 28
N
7 = constant
Since the specific heats at station 8 and 9 are approximately equal
(since T8 -- T9s ), we may assume a common specific heat cp . In this
case, the theta exponent x reduces to 0 (since Ah = c p Al). Thus, we

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

Corrected Torque: Q / S
arrive at the classical correction for thrust, namely
The correction for torque follows from corrections for speed and
horsepower, since horesepower is the product of torque and speed. It
" 62 follows that a corrected horsepower should be a product of corrected
torque and corrected speed, i.e.

Corrected Horsepower: HP / S HP* =Q*N*

Horsepower required (or developed) by a compressor (or turbine) is =_Q
HP _ * N
given by

HP = -.-W
W. c Ah L W c AT where J = 778.17 ft lb / BTU _HPQ
550 p 550 '
NS S
A corrected horsepower will not vary with inlet temperature and
pressure and thus can be written as above with corrected flow and Corrected Acceleration Ndot: N / S
temperature, i.e.
From Newton's Law, Torque (force) = Moment of Inertia (I) x
Acceleration:
HP*= J W *c Ah*^ _ W *c AT*
550 W. ° 550 ° p Q=IN
_ J W. JO cA
_ T
p -- Q*=1N* s =1 N*
550 S 8
-1 IV
p AT
-_ 550 W° c
S^ N-N/6
HP
SJ
Corrected Metal Temperature Rate Tm dot: m
e .74 5 .8
Corrected Rotational Speeds: N /
Tangential velocity is related to rotational speed (rpm) by the radius The (time) rate of change of a metal temperature is typically modeled
of the object (compressor or turbine blade) in question. It is also as a 1st order lag heat transfer between the gas path temperature T and
related to acoustic velocity by Mach number and the square root of the metal temperature Tm . In symbols,
temperature.
Mathematically, we have
Tm =T(T-T,n )

v=rN=M„ 9- T me (3)
where time constant ti = p
HA
where v is the velocity in ftls, r is the radius (ft), N is the rotational
speed (rpm), g, k, and R are constants (gravity, ratio of specific heats The McAdams correlation for turbulent flow over a flat plate
and the gas constant respectively), and Mn is Mach number. Taking (Shepherd, p254) is given by
logs and differentiating, we obtain
Nu = 0.023(Re) o.s (R,) 0.4

7
 where Nu, Re, and Pr are the Nusselt's, Reynold's and Prandtl's Tm * H * C p T* -Tm *
numbers respectively which are defined as follows:
7m H J c P * T Tm
HD
Nu = Re = v
p D , Pr = -----
0.4-0.2„ s -o.s
K ^.t K
=6
T-T.
-0.6-0.2n 8-0.8
Substituting and re-arranging terms we can obtain the following form:

Tm

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

H= 0.023 ' ( PvD °8
)
m = e 0.6+0.2n a
C 0.8

K 06 C 4 p 0.8 v 0.8
= 0.023 D oz µ0.4 The value of n appearing in the theta exponent can be determined
experimentally by correlating viscosity with temperature from which
cPµo.2po.8vo:8 it is observed that n varies from 0.8 at low temperature to 0.6 at high
_ 0.023 K0.6
C 0.6 4 0.6 D02 temperature. These values provide a theta exponent ranging from 0.76
P to 0.72, respectively. The average value of 0.74 is recommended, thus
023 ) c µo.2po.8vo.8
= ( ^O.
providing the requisite correction.
l10.6
P D0.2
cPµ 0.2 p 0.8 v 0.8 T.
m a0.74 80.8
0.027
D02

(4)
DETERMINING CORRECTION FACTORS
where the final approximation is obtained by noting that Prandtl's
The methods for obtaining appropriate corrections for gas path
number is approximately constant in the range of interest (Shepherd, p
254). At this juncture we may recall that mass flow Wa = p A v was parameters depends in part on the application at hand as well as the
amount and type of information (data) available. These methods can
corrected by square root of theta over delta. Combining these
observations, we have be split into essentially two groups: 1) analytical and 2) empirical. In
practice it is more than likely that a combination of both types of
techniques will be utilized. We will briefly illustrate both of these
WQ - WQ*6 -pv approaches.
A
Analytical Methods
Therefore, As a practical application of how to utilize some of the above
= (WQ *) o.8 6 0.8 0 -0.4 information to determine a theta correction for Wf in a turbo-fan
(pv)0.8
engine from engine simulation information, consider the information
contained in Table 2. This data was obtained from a non-linear aero-
Using the approximation at the end of equations (4) to define the thermodynamic representation of a small turbofan engine. In order to
reference condition H* and noting that D=D*, we can write the ratio avoid introducing new station numbers, we will use the
corresponding station numbers of our sample engine in Figure 1. The
H / c* ^ ( 4µ*) 0.2/_ v *\ o.s thermodynamic deck provides the following information at station 4
(HPC exit) and station 5 (HPT inlet) at one engine operating point.

__C C c*J en) 0.2 (_ro s0.8 Applying these values to the Wf Equation (5) yields a value of 0.707
for the theta exponent `x". The actual value used for this engine is
0.69. Presumably, the above calculation would yield slightly different
results at differing engine operating points, the average of which is
= CP e0.4-0.2n 5-0.8
probably around 0.69.
cP

where we have assumed that the ratio p*/µ is solely a function of

temperature (6 ). Now using equation (3) to define Tm * in a similar
manner, we can form the appropriate ratio and use the above
relationship to simplify the expression.

8
 Table 2: Non-linear Engine Simulation Data 0.5. All of these theta exponents derivable from a differential model
will provide a correction that is closer to the truth than the classical
Parameter Station 4 Station 5 Units (providing the model is reasonably correct) and can be used as a good
first guess for the correction to be used. The fidelity of these
k 1.39801 1.30214 ft-lb/deg R
corrections will of course depend on the fidelity of the underlying
h 300.297 657.18 BTU differential model. For general analysis, classical corrections often
T 1235.18 2436.52 deg R suffice. Several parameters may require a finer correction. Wf is
typically one of those parameters and it is not uncommon to see theta
Wa 23.2613 23.7252 pps exponents ranging from 0.5 to 0.7 for this parameter. Such a
Wf 0.4639 pps refinement is required for work demanding greater accuracies, such as

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

diagnostics, and for these purposes it is customary to use corrections
Ah 356.883 such as those given in the sensitivity matrix above, and when actual
WI/Wa 0.019943 engine data becomes available, to adjust these first guesses
empirically. This leads us to our next topic.
cp 0.24081 0.29542
cp T / dh 0.83346 2.01690
1 +Wf/ Wa 1.019943 Table 3: Sensitivity Matrix
TAM PAM
Exponent 0.707
NIC2 -0.0105 0.0077
N2C2 -0.0167 0.0021
WF 0.6014 0.9933
This type of analytic approach could be made by applying some of the FN 0.0123 1.0035
intermediate equations used in the derivations for various gas path
P3/P2 -0.0119 0.007
parameters in much the same way as in the above example. Another
analytic method would be to construct a complete differential model P4/P3 -0.014 -0.003
of the engine [Volponi and Urban, 1992]. Models of this type provide P7/P2 0 0
sensitivities to % changes in independent parameters (T2 is one such) T3 0.9816 0.0037
to corresponding % changes in dependent engine parameters (such as
T4 0.9291 0.0025
Wf, for example). The difference between this and what was done in
the derivation of the Wf correction, is that the latter would be one of T5 0.9265 -0.005
many differential equations (usually around 40) that would relate all T6 0.9542 -0.005
the independent and dependent parameters at various engine stations T7 0.9746 -0.006
interrelated through the laws of thermodynamics, energy and flow P3 -0.0119 1.007
conservation. The simultaneous solution of these 40 equations would
P4 -0.0259 1.0037
produce (for example) a sensitivity of T2 on spool speeds ( N ), fuel
flow ( Wf), temperatures at various stations, etc. These sensitivities P5 -0.0224 1.0031
give an indication of the (theta) correction to be applied. P6 -0.0073 1.0024
WA -0.5076 1.0056
An example of such an approach is illustrated in the sensitivity matrix T2 1 0
in Table 3 below.
P2 0 1

The sensitivities of engine inlet temperature and pressure are given in

the last two columns for all of the dependent parameters listed in the
1st column. For example, the dependent parameter NIC2 which is
corrected low spool speed has a sensitivity of -0.0105 with Tam. This
means that the following is true: Empirical Methods
Empirical methods utilize actual engine data collected over a
temperature range and comprise more of a statistical approach to the
dNlc2 dNl _ 1 dT2 dT2 dNl dT2 subject of parameter normalization and as such, do not require any in-
° -0.0105 T2 Ni -0.4895 T2 = 0 depth knowledge of gas turbines or their operation. It has become a
Nlc2 Ni 2 T2 generally accepted practice among engine performance diagnosticians
to perform some sort of empirical analysis to verify the validity of the
which implies that the correction for Ni is Nile. 4895 instead of correction factors being used. Neglecting to do so exposes the
Nl/60. 5 . Likewise, the correction for N2 is .4833 instead of square diagnostic procedure to greater error which may render the
root. Now if we choose some raw parameters such as T3, T4, and T7, diagnostics entirely useless.
we see that the suggested theta exponents are 0.9816, 0.9291, and
0.9746, respectively, and not 1.0 as the classical correction would
dictate. Fuel flow is demanding an exponent of 0.6014, instead of
There are many techniques that can be put forward to analyze engine
data for normalization purposes, however, we will restrict our m(8 —1)
discussion to a simple formula (derived by this author, circa 1980), xza+ ( (6)
100ln(8)
suggest several alternative approaches, and finally offer some Monte
Carlo simulation results that suggest that all of the above produce
roughly the same result. We see from this equation, that if the slope of the regression line is
positive, then the exponent will be raised. Likewise, it will be lowered
For simplicity, we will consider only temperature (8) corrections; when the slope is negative. If the data sample was such that the mean
pressure (6) corrections can be handled in an analogous manner. Let theta was standard day (i.e., =1), then note that the limit of (8-1)/ln(8)
P*=P, where x denotes the true (sought) exponent and let a be a as 0 —+ 1, is 1, and the equation reduces to

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

theta exponent used to normalize a set of data {Pi }i=1,...,n• Let us
further assume that each datum Pi (for all i) was obtained at an inlet
x -- a + 100 (7)
condition represented by 8i and that we have available, the nominal
value of this parameter for the engine operating condition(s) at which
the parameters Pi were obtained. Let us denote this uncorrected The form of these equations is particularly simple and allows a rapid
nominal value by PNOM . Then check for corrections when engine data is available. The reader should

(
note that the above can be applied just as well to pressure corrections
, AGi (i.e. the 6 exponent) where the relationships would be identical except
8a 8x PNOM 100 I )
for replacing 0 with 8).

There are, of course, other empirical paths which one can take in
where OGi represents the % of point delta from nominal (for all i).
pursuit of the proper correction. For instance, one could determine
Now if we plot these deltas against 0, then if a were the true
the exponent which minimizes the standard deviation (aAG) of the
correction exponent, we would expect to see a scatter plot that would
parameter deltas {AGi}, or to minimize the covariance
have zero correlation, i.e. a linear regression line fit horizontal with
(GAG,9)between the deltas {AGi} and {81}, or the correlation
the theta axis. An upward or downward trend, however, would
coefficient (r,G 8). Monte Carlo simulations indicate, however, that
indicate the extent to which the exponent a was deficient. If we let m
the results are approximately the same for all of these techniques. The
and b, denote the regression slope and intercept respectively, then we
differences obtained for theta exponents using these four methods are
have a defined linear relationship as follows:
well within the limits of measurement noise and are not statistically
significant. The three minimization methods which we have just
AG=m9+b where b=tG—m8 alluded to, do carry a substantial computational overhead in that each
of these methods are iterative in nature and require that a complete
!AG. _m[ ti]
!
statistical analysis be performed at each newly chosen exponent,
where the exponent is modulated in small increments over some
range. After that is completed, the minimum statistic is isolated and
and m=r SAG the corresponding exponent is selected as the winner. In this regard,
se
the simple formula(s) given in Eqs. (6) and (7) above are superior in
that they return essentially the same exponent but require only one
where r is the correlation coefficient and s4G and s8 are the sample statistical analysis.
standard deviations for AG and 8, respectively. Thus for arbitrary 0,
we have As an illustration, Table 4 below presents the results of a Monte Carlo
simulation comparison of the slope formula method and the minimum
standard deviation method. This table contains 50 randomly chosen
true exponents between 0.7 and 1.0. For each exponent a set of 200
deltas {AGi} were randomly generated. These deltas were selected
P P 8" ° 1+ 1+- (9=1) from a normal population where the variance for each of the 50 {AGi}
PP
— ---

NOM m +b NOM 100

sets was also selected randomly. Thus, we have created a set of data
m8+b where we have differing true exponent magnitudes and as well as
and P 8"_" ^1+ (8=8)
PNOM 100 differing variation in the deltas (simulating different measurement
therefore accuracies and/or engine to engine variation levels). For each of these
50 true exponents sets, we set a fixed corrected parameter level P*
m8+b l+ m+b and generate 200 8i's and calculate a set of 200 un-corrected
x-° l+
100 j L 100 J parameters {Pi} by applying both the theta correction (in reverse)
using the true exponent, as well as the % delta {OGi}. This produces a
Taking natural logs of both sides and applying the relationship that set of data that simulates 200 raw measurements, each at a different
In( l+x ) z x for small x, we obtain an expression for the new temperature, each with a different (measurement/engine) bias ({G}).
exponent in terms of the old exponent, namely, The {Pi} are now corrected using one of 50 randomly chosen
The
incorrect reduction theta exponents. Using this set of falsely corrected
parameters, the aforementioned 2 methods are applied and an estimate method, however, offers the additional advantage of requiring far less
for the true exponent determined. This is repeated for each of the 50 computational effort
samples exponents. Table 4 contains a sampling of the data generated.
Figure 3 below summarizes the results of all 50 sets in graphical form. SUMMARY
We have summarized the traditional standard day corrections for most
Table 4: Monte Carlo Simulation of the commonly encountered gas path parameters such as
temperatures, pressures, speeds and flows as well as providing a
Min Std Deviation derivation for each utilizing only simple thermodynamic relationships
Slope Method Method and fundamentals of the calculus. The topic of dimensional analysis
TRUE Reduction was also mentioned as a matter of historical perspective in the

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1998/78668/V005T15A026/4218890/v005t15a026-98-gt-347.pdf by guest on 02 May 2024

Exp Exp Exp Error Exp Error treatment of non-dimensional quantities and their relationship to
0.710 0.789 0.707 -0.0035 0.705 -0.005 corrected parameters. Other model dependent analytical methods were
0.702 0.899 0.710 0.0085 0.710 0.0082 also discussed and illustrated. Finally, and perhaps most importantly,
0.953 0.756 0.954 0.0008 0.950 -0.0028 the practical aspects of determining the theta-delta exponents
0.773 0.864 0.768 -0.0046 0.775 0.0025 empirically from samples of engine data was discussed and a
0.866 0.827 0.866 0.0006 0.865 -0.0005 computationally simple method was derived and illustrated.
0.964 0.710 0.986 0.0218 0.980 0.0158
0.743 0.936 0.734 -0.0086 0.735 -0.0075 ACKNOWLEDGMENT
0.749 0.706 0.763 0.0135 0.760 0.0109 The author wishes to thank Richard Meisner formerly of Pratt &
0.953 0.701 0.950 -0.0023 0.950 -0.0025 Whitney Aircraft for his insightful critique of this paper as well as for
0.708 0.809 0.711 0.0027 0.715 0.0071 the many hours of phone conversations, communications and
0.791 0.787 0.784 -0.0072 0.785 -0.0063 suggestions relating to this general topic. I would also like to
0.961 0.747 0.956 -0.0045 0.960 -0.0008 acknowledge my long time friend and colleague Louis A. Urban, for
0.775 0.893 0.738 -0.0366 0.740 -0.0347 introducing me to my first corrected parameter.
0.860 0.934 0.850 -0.0098 0.855 -0.0052
REFERENCES
Buckingham, E., 1914, On Physically Similar Systems: Illustrations
0.920 0.988 0.913 -0.0079 0.915 -0.0054 of the Use of Dimensional Equations, Physical Review, Vol 4.
0.729 0.703 0.727 -0.0017 0.725 -0.0039
0.821 0.729 0.844 0.0223 0.845 0.0237 Capon, R. S., and Brooke, G.V., 1930, The Application of
0.867 0.972 0.862 -0.005 0.865 -0.0018 Dimensional Relationships to Air Compressors, Aer. Res. Comm.,
0.774 0.943 0.770 -0.0039 0.770 -0.0035 Reports and Memoranda No. 1336.
0.864 0.811 0.864 -0.0007 0.865 0.0007
0.927 0.711 0.929 0.0021 0.930 0.003 Sanders, N. D., 1946, Performance Parameters for Jet-Propulsion
0.978 0.986 0.976 -0.0017 0.975 -0.0029 Engines, NACA Technical Note No. 1106. -
0.756 0.796 0.757 0.0004 0.755 -0.0012
Shepherd, D.G., 1949, An Introduction to the Gas Turbine,
Mean Error = -0.0011 -0.00053 Constable and Company, Ltd., London.

Volponi, A. J., and Urban, L. A., 1992, Mathematical Methods of

eo Relative Engine Performance Diagnostics, SAE 1992 Transactions,
0.05 Vol. 101, Journal of Aerospace, Technical Paper 922048.
0.9
o m
Warner, D.F., and Auyer, E.L., 1945, Contemporary Jet-Propulsion
0.85 O"
Gas Turbines for Aircraft, Mech. Eng., Vol 67, No. 11, 1945, pp.
0.9
; o B oa; 707-714.
o
EsEm .70
0.75 BOO

0.7 _Y______________________^____________-_ o SMDWM*.

pO ^ ^ •R^pwbn SNP I.MOnE ^ _

0.05
0.7 0.75 0.8 0.0 0.9 0%

Figure 3: Monte Carlo Simulation Experiment

As can be seen by perusing the table, a reasonably accurate

assessment of the proper exponent can be obtained using the empirical
method described above. In addition, both methods used in this
example return exponents which are essentially the same. The slope