THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

345 E. 47th St., New York, N.Y. 10017 96-GT-164

The Society shall not be responsible for statements or opinions advanced in papers or discussion at meetings of the Society or of its Divisions or
Sections, or printed In its publications. Discussion is printed only if the paper is published in an ASME Journal. Authorization to photocopy .
material for internal or personal use under circumstance not falling within the fair use provisions of the Copyright Act is granted by ASME to
libraries and other users registered with the Copyright Clearance Center (GCC) Transactional Reporting Service provided that the base fee of
$0.30 per page is paid thirty to the CCC, 27 Congress Street, Salem MA 01970. Requests for special permission or bulk reproduction should be ad-
dressed to the ASME Tectmical Publishing Department

Copyright 0 1996 by ASME All Rights Reserved Printed in U.S.A1

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

HOW TO GET COMPONENT MAPS FOR AIRCRAFT
GAS TURBINE PERFORMANCE CALCULATIONS

Joachim Kurzke 111111111111, t11111111111

Engine Performance Department
MTU Munchen GmbH
Miinchen, Germany

ABSTRACT
Precise simulations of gas turbine performance cannot be done without circumferential speed
component maps. In the early days of a new project one often has to use velocity
scaled maps of similar machines. Alternatively one can calculate the mass flow
component partload characteristics provided that the many details needed
for such an exercise am available. In a later stage often rig tests will be done a auxiliary coordinate
to get detailed information about the behavior of the compressors isentropic exponent
respectively turbines. 11 efficiency
Performance calculation programs usually require the map data in a density
specific format. To produce this format needs some preprocessing.
Measured data cannot be used directly because they show a scatter and they
are not evenly disnibuted over the range of interest. Due to limitations in the INTRODUCTION
test equipment often there is lack of data for very low and very high speed. For off-design gas turbine performance calculations one needs a good
With the help of a specialized drawing program available on a PC one numerical representation of the component behavior. Remember that
can easily eliminate the scatter in the data and also inter- and extrapolate typically 1% loss in compressor efficiency translates into 0.5% increase of
additional lines of constant corrected speed. Many graphs showing both the specific fuel consumption. This is not a negligible number. It can make the
measured data and the lines passing through the data as a function of difference between an acceptable and an unacceptable engine design.
physically meaningful parameters allow to check whether the result makes The preparation of component characteristics for performance
sense or not. The extrapolation of compressor maps toward very low speed, calculations can be a cumbersome task. When they are based on measured
as required for the calculation of starting, idle and windmilling performance data, then during rig tests often not the full range of interest is covered and
calculations, is discussed in some detail. extrapolations are required. Moreover, especially at low speeds where the
Instead of true measured data one can use data read from maps scatter in the data can be significant, it is not easy to interpret the data
published in open literature. The program is also an excellent tool for correctly.
checking and extending component maps one has derived from sparse One can attempt to smooth, inter- and extrapolate the data with a
information about a gas turbine to be simulated. computer program automatically. However, such a program has to be very
sophisticated because it is not sufficient, to apply some mathematical routine
which finds the best approximation of an arbitrary function z-f(x, y) to the
NOMENCLATURE scattered data In a compressor map, for example, the speed lines are mostly
A area vertical in a part of the map and this will impose some restrictions for the
specific work function z=ftx, y). How to tell the program, where the speed lines become
Mn Mach number vertical, especially, when them are no or only a few measured data in this
spool speed region?
gas constant Another problem arises with variable geometry on compressors.
total pressure Normally one can assume, that transitions from one speed line to the next
P, static pressure are smooth. However, when the schedule for the variable geometry setting
total temperature - f(NNT) has a sharp kink because the mechanism hits a mechanical
T, static temperature stop, then this will be seen in the correlation between flow and speed. The

Presented at the International Gas Turbine and Aerotngine Congress & Exhibition
Birmingham. UK — June 10-13, 1996
a utomatic smoothing program must not remove the kink in the function
W—f(NATT) in this case
Certainly one can write a fully automatic map preparing program, but W -A (y, R, At)
his will not be a trivial task for some of the reasons explained above. An
alternative approach is, not to do everything automatically but to leave the

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

engineer in the loop. He who knows how a compressor map must look like,
can decide best about drawing curves through the measured data. He can The circumferential speed U expressed as a Mach number is
deal easily with outliers, kinks in schedules, interpolation and extrapolation
etc. The task of the map smoothing program is to provide the engineer with -c (y, R, Mn)
convenient drawing tool and many graphics which allow to judge whether /ITT, it
the lines drawn make physically sense or not.
SMOOTH C and SMOOTH T (acronyms for smooth compressor and
smooth turbine map) are semiautomatic PC programs that give the engineer Specific work H has the dimension of speed' and can therefore be
full freedom forithe interpretation of his data. From menus one can select. normalized by dividing through yRT,:
many graphs with physically meaningful correlations. Modifications of the •
lines drawn through the scattered points are done easily with the mouse. H H
Interpolation and extrapolation of maps are possible with good quality y R T, T
herimot a lot of correlations can be checked. The results are consistent and.
physically Meaningful component maps of high quality as required for
Precise gas turbine performance calculations. For fixed isentropic exponent y and gas constant R the three quantities
WIT/P, NA/T and H/T define all Mach numbers throughout the turbo
machine provided that the geometry remains unchanged. This implies also,
MAP FORMAT that all flow angles are given with these quantities. Since the losses of blade
Nearly every gas turbine performance calculation program uses its own and vane rows depend primarily on incidence angle and Mach number, also
data format for the component map representation. From a mathematical the efficiency of the turbo machine is fixed. From efficiency and HIT one
point of view there are advantages for specific data arrangements. As long can easily calculate the pressure ratio.
as When the Reynolds number is low, the losses and, thus, the efficiency,
.1 the physical content of the data is the same, it is only of secondary
importance which format is used. One can easily convert maps using depend not only on Mach number and incidence but also on Reynolds
specific work as parameter into maps with pressure ratio, for example. The number. This effect is normally dealt with as a correction to the map. In the
following text the precise terms "corrected speed N/VT", "corrected mass
d ata arrangement has an influence on the calculation time, on the amount
oi f data to be stored and on the convergence behavior of the performance flow WIT/P" and "corrected specific work HIT" are mostly abbreviated as
program. However, when the adequate amount of data is stored, then the speed, flow and work.
result of the calculation will not depend on the data format
I For evaluating a map one normally needs to interpolate between the
tabulated data. This can be done linearly or in a more sophisticated manner.
Compressor
In a standard compressor map there are lines of pressure ratio P,/P, over
The linear interpolation has the advantages of being fast and always yielding corrected flow WVT,/P, for several values of constant corrected speed
a reasonable result while the quadratic interpolation, for example, can be NWT, as well as lines of constant efficiency n. Such a "single map" is an
more accurate. However, the latter is true only if the data field is smooth. appropriate description of the performance of gas generator compressors.
When the data contains some scatter, then a nonlinear interpolation can For the fan of a low bypass engine, however, one needs a more complex
yield surprising results. Another argument for linear interpolation is, that map format which is called a "split map." This is because the flow is split
there might be physically justified discontinuities in the data. These can be between the core and the bypass downstream of the fan and there are
introduced in a compressor map, for example, when the variable geometry significant differences in pressure ratio and efficiency between both streams.
setting is not a smooth function of corrected speed.

Simile Map
General Considerations A compressor map in the standard format cannot be used directly in a
The basis for the standard format of every turbo machinery map is the performance calculation program. It is not possible to read efficiency from
Mach number similarity and from principle all parameters (except such a map with given speed NAIT, and flow WIT,/P1 because in a part of
efficiency) in such a map are equivalent Mach numbers. Instead of the mass the map the speed lines can be vertical. It is also not possible to read the map
flow W the corrected mass flow WVT/P is used which represents the Mach with given speed NWT, and pressure ratio P,/P, because it might be, that
number at the inlet of the machine. Conservation of mass W=ApV can be there are two values for corrected flow at a given pressure ratio PIP, (see
expanded as follows figure 1).
The solution to the problem is to introduce auxiliary coordinates (here
W=A V called B-lines) which intersect the speed lines. This allows to read the map
I? Ts independent from the shape of the lines with given 0 and speed NWT,.
From principle the &grid can be selected arbitrarily as long as there are
_ P A P,
P.\17 \r
y unambiguous intersections of the 8-lines with the speed lines. However, a
clever definition of the 0-grid can give B some significance from a
fi T, 0
1,.r1 R compressor aerodynamics point of view. In SMOOTH_C the &lines are
equally spaced parabolas with parameter values of 0 = 0...1. It will be

2
discussed later how B-lines defined in such a way can be used to draw Split Map
important conclusions.
It is left to the user of the program, to define the position and the shape In the case of fans for high bypass ratio enginei it is common practice,
of the limiting parabolas 13-0 and 0-1. It is prudent, to let the upper parabola to use for the booster and the bypass stream separate ("single') maps. For
(0-1) pass through the point (W/T,/P, -0, P 2/P 1 =1} and to set it slightly fans of engines with a low bypass ratio often so-called "split maps" are used.

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

In a split map thereare five lata . for each measured point the total
compressor mass flow WiT i/P,, the two presiure ratios PIP, (bypass
stream) and P21/P, (core stream) and the two efficiency values n w and Thai.
The B-grid can be defined in one of the graphs P 1j/P 1=f(NA/71 , WiT1/P1 )
or PIP,-f(NaT,, WM 1/1'0, dependent on the.nominal bypass ratio.
When this is less than one then the definition P21/131 =1(NA1T 1 , Wira,)
should be preferred and otherwise the alternative definition.
The map in which the 0-grid is defined is called the primary map and is
dealt with like a "single" compressor map. When the speed and efficiency.
lines for the primary map have been drawn through the data points then in
the next step the secondary map is produced by an indirect method. The
data of the secondary map are presented as profile factors that are defined
as

11.13

11.21

_ PI IPI - P13
JP/P p
Figure 1: The compressor map reading problem 21 1P 1 P21

above the surge line. The lower parabola (0=0) should be roughly parallel
to the line (B- I) in such a way, that the two lines do not cross each other
within the range of interest A proper definition of the B-grid will also result
in the line 0.0.5 being roughly parallel to the peak efficiency line over most
of the speed range. Figure 2 shows a 0-grid overlay with measured data
points and the peak efficiency line, for example.

Figure 3: Pressure profile of a modem fan

In the plots [0, NA/TO and fm, - f (B, NAITO, in which the data for
the profile factors are shown, there are again smooth lines drawn. Pressure
ratio and efficiency for the secondary map can be derived easily from the
Figure 2: Beta-line grid primary map and the profile factors. An example for the pressure profile
factor of a modem three-stage fan is shown in figure 3.
When the 13-grid is defined, then in a first step the smooth lines through
the scattered data of P T/P,4(WiT 1/P 1) are drawn for each speed NWT, One may ask, why the secondary map is not produced in the same way
("speed lines"). After that the efficiency data will be shown as as the primary map. This is not possible because in the secondary map the
naf(WiT,/P,) or trf(P,/P0 and again smooth lines are drawn. For this 0-lines are not parabolas in the plane P„/P, - fiNaT,, W11 1/P0 [in
second step one can also switch to graphs with Flacr-f(W/T 1/P0 and contrary to the B-fines in the primary map F21/1 3 NA/T1, WiTI/P,)] since
WIi -MYPO. the pressure profile factor PIP„ is not constant

3
Turbine minimum number of data. The map format with 13-lines has also a numerical
For turbines there are also a lot of different map formats in use. In gas advantage over the often used tabulation of ri—f(H/T I , NIT). Reading data
turbine performance calculation programs often corrected speed NWT, and at an arbitrary NWT, value needs an interpolation which for the reasons
pressure ratio PR I or specific work WT I are known when the turbine map discussed above is often done linearly.
is to be read during the calculation. With these quantities given there is no

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

map reading problem like in the case of compressors. One can
unambiguously evaluate the map and get both efficiency n and corrected
flow WIT/P1 . Nevertheless it is convenient to use in turbine maps auxiliary
coordinates, i.e., 13-Imes. Look at figure 4 which shows the turbine
characteristic of a high speed three-stage research turbine as pressure ratio Interpolation

Pressure Retie or Spec. Wor k

over corrected speed with lines of constant efficiency. Toward low speed the Error

•
region with optimum efficiency becomes very narrow. When one needs to
accurately describe the low speed part of the map, then the step size for the
low pressure ratio values stored in memory must be very small. However,

016%.
small pressure ratio steps are not needed for high corrected speeds.
Moreover, the high pressure ratio values required for the description of
turbine performance near to nominal speed are not interesting at the low
speed end and to store such data is unnecessary.
One way out of the problem is, to use a 8-grid where 13=0 yields
f (N/VT I) and Del is defined as (P 1/132).= f (NA/T,), see figure
N N
4 All the relevant parts of the map will be enclosed between these two lines.
M7 7' 7f
At low speed the high pressure ratio values are normally irrelevant and the
d stance between the lines for 6=3 and 6=1 can be small. At high speed
there is no big change of efficiency with pressure ratio and it is appropriate,
to tabulate big steps in pressure ratio. Thus, one can achieve an accurate Figure 5: Linear interpolation error
description of the turbine efficiency over the full speed range with a •

Corrected Speed
Figure 4: Turbine map with lines 13=3 and 13=1

4
 As shown in figure 5 the map reading error due to linear interpolation excellent tool to check, whether inter- and extrapolated speed lines fit to the
can be significant when the distance between tabulated Hrr, values is big. rest of the map or not.
This is because H/T for peak efficiency varies with corrected speed. With
the 0-line grid one can achieve, that 0 for peak efficiency is nearly Losses
independent from corrected speed and thus the error due to lineau One can see from a compressor map plot with "efficiency islands"

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

interpolation is minimized. whether the map data am smooth or not. However, one cannot see whether

CORRELATIONS
There is an old Latin wisdom which states: "nature non fecit saltas"
(nature does not jump). This holds also for the normal operating range of
turbo machines, where the flow pattern changes gradually from subsonic
over transonic to supersonic with increasing speed. Therefore all aero-
thermodynamic parameters in the map of a turbo machine will be smooth
when plotted over speed or 0 except for the effects of discontinuous variable
geometry settings or interstage bleed flow. 1
Compressors

Flow - Speed
The 0-lines are smooth and therefore the corrected mass flow plotted
over speed must also be smooth for 0—constant. Figure 6 shows the mass Speed'
flow characteristic of a three-stage research fan as a typical example. The
line for 8.:0 (i.e., lowest pressure ratio) represents the highest mass flow for Figure 7: Specific work as function of 0 over speed' for a three-stage fan
each speed. While in the low speed range the mass flow increases
progressively with speed, this is no longer true at the top end. There the the data makes sense physically. For that purpose it is much better to look
compressor chokes and the flow increase with speed flattens out. at the shape of the lines (H - Fl i,,„„)/N7 f (H/N'). These lines look like a
bucket with a consistent trend of the minimum loss location with speed (fig.
8). When the compressor exit chokes, then the work done remains constant
and consequently the left part of the "bucket" must be a vertical line. This
is quite normal for radial compressors with a supersonic diffusor inlet flow.

Extrapolation to very low speed

For the calculation of engine starting and windrnilling one needs
component maps that include performance at very low speed. The
conventional compressor parameters like flow coefficient V./U and work
coefficient H/U2 as well as efficiency obviously cannot be used when
N/i1-1-0 and are of no help for map extrapolation. However, some simple
considerations lead to important conclusions: When the compressor is not

Speed
111111111111111111
Figure 6: Corrected entry flow as function of 0 and speed
Pra11111111112111
•1111•111? 40,1/4
When the line lint is selected to pass through (W/T 1/P1 -0,PiP1..1),
then the line 0-1 must pass through (WIT,/P 1 -0, N11T1-0) in the plot
flow over speed. This is a help for extrapolating the map toward lower to
speeds.
Illitirati
nark:1mA
The work done in a compressor is proportional to the square of
circumferential speed. Therefore it is convenient, to plot specific work HIT,
over (NWT,)2 for constant B. Since at zero speed definitely no work is done,
all 0-lines must pass through the point (WfrO,NWT,=0) independently
mismanran Specific Work! Speed'

from the definition of the 0-grid. As can be seen from figure 7, which was Figure 8: Loss characteristic of a five stage axial compressor
derived from a modern three stage fan, the 0-lines are fairly straight and an

5
turning at all, then no work is done, the pressure ratio will be equal to or Losses
lower than one and the mass flow can be positive or negative. The speed When a turbine map is plotted in the same format as a compressor map
line N/VT,=0 will have a horizontal tangent at {W./T,/P,I, PJP,-1}. The (i.e., with pressure ratio over flow for several values of corrected speed)
peak efficiency for speed 1 RPM (just an epsilon above zero) must be also then one encounters a problem. In the region around the design point the
at {WiT1/P,:),Pi/P,-1} because obviously there are the smallest losses. speed lines are very near to each other or do even collapse and obviously
A test in which the pressure loss of a compressor with a fixed rotor is one cannot plot lines of constant efficiency in this type of graphic. Therefore

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

measured would be extremely useful for setting up a map which is suited turbine maps are shown as pressure ratio (or specific work) over corrected
for engine starting and windmill simulations. Normally, such a test is not speed as figure 4 or over the product of corrected speed and corrected flow
feasible and one has to estimate from theory the pressure loss of a W/TIP,* NWT,.
nonrotating compressor. The efficiency contours in such a map must be smooth for a good
Between the speed line zero and the normally measured speed range one representation of the real turbine behavior. However, one cannot judge very
can do fairly good interpolations with a program like SMOOTH_C. In this
part of the map one will encounter the problem that efficiency jumps from
to +in when the effective specific work H/T, passes from positive to
negative values, see figure 9.
1%I

( Correct Interpretation
Efficiency _ H iserS
Heftective

0.61

Sp. Work
1.2 14 16 18 20 2.2
Pressure Ratio
Flow

Pressure Ratio = 1
Figure 10: Different interpretations of real turbine test data
well from the shape of the efficiency islands whether the data is physically
meaningful or not. For this purpose another format is more useful where
Figure 9: Efficiency and specific work for low speed conditions efficiency is plotted over the turbine velocity ratio NI./T/'/(2H_ „ / T)

In the region with pressure ratio below one the map can be checked, and
if necessary, corrected in the plot with H/I,=f(WiT,/P,, WT I). There also
the line for Hsi, is shown which must always be below the line for H. at 0. iv, - 2%1
otherwise the second law of thermodynamics would be violated.
In the normal compressor operating range with pressure ratio higher than
one at least the approximate location of the peak efficiency point can be
found by interpolation from the bucket point in the loss characteristic
(H-1-1,,n)/N2 = f (H/N2) (see fig. 8) and the fact that at zero flow the losses .015
have a minimum for speed "epsilon". The level of the peak efficiency has 0.89
to be extrapolated from the loss characteristic and is certainly not zero.
0.81
0.61 0.71
Turbines
Incorrect Interpretation)
Flow 0.07
0.03 0.04 0.05 0.06
It depends on the design of the turbine, whether the maximum corrected Turbine Velocity Ratio
flow Wit/P1 is dependent on corrected speed NUT, or not. Highly loaded
one stage turbines do often operate at design pressure ratio with a choked Figure 11: Turbine efficiency as function of corrected speed and turbine
stator and in this case obviously the maximum corrected flow remains velocity ratio NaT/(2•11i,n/T)
constant in a significant part of the map. When the (first) stator of a turbine
is not choked, then maximum corrected flow becomes a weak function of
corrected speed (W,/T,/P, decreases when NWT, increases). Figure 10 shows an example with real test data and two different lines.
It is general practice to plot the corrected flow over pressure ratio or The dotted line passes nicely through the measured points and seems to be
specific work. In a plot over speed with 0-lines one can check the a very good interpretation of the data. However, when this line is drawn into
consistency of trends both for maximum flow and the partload range. the graph with efficiency as a function of turbine velocity ratio (figure 11)
it becomes obvious, that the line is not correct. From comparisons with the

6
neighborhood and the knowledge about the basic shape of the curves one the incorrect data instead of going back to the map calculation program and
can derive a better interpretation of the data. try to solve the convergence problem.
Gas turbine performance calculations are done at many universities,
academies and research institutes. Often these institutions do not have direct
APPUCATION access to component test data and they cannot calculate the maps they need.

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1996/78767/V005T16A001/4465067/v005t16a001-96-gt-164.pdf by Valparaiso University, Hamid Chenarani on 31 July 2024

Originally the programs SMOOTH_C and SMOOTH_T were They have to rely on maps that are published in open literature. The figures
developed for use with rig test data. It has been seen later, that they are also in Many publications are tiny, have only a limited number of efficiency lines
an excellent tool for checking component maps that have been prepared for and sometimes even typing errors. Nevertheless, one can blow the figure up,
performance calculations by other means. Inconsistencies and physically read the date as precisely as possible and use it as "measured data" for input
incorrect extrapolations, especially toward the low speed region, are into SMOOTH_C or SMOOTH_T.
immediately obvious from the many cross plots available. This was done with figure 12 which was published in ref. I in the size
as shown here. The design efficiency was arbitrarily assumed to be 80%
which will not be far from the truth. Figure 13 shows the smooth map with
additional inter- and extrapolated efliciency.values. Note that the speed line
designated as 105% in the original is marked as 107% in figure 13 because --
then this line fits much better to the rest of the map in the flow - speed and
the work - speeds plots. The more important output of the program . is the
tabulated data which can be used with the PC program GASTHR.B (ref.2)
which is available at many universities or with any other gas turbine
performance calculation program.

REFERENCES
[I] D.L.Palmer, W.F.Waterman
Design and Development of an Advanced Two-Stage Centrifugal
2.0 3.0 4.0 50 6.0 7.0 130
Compressor
Comocrad Akflow5 Ids ASME 94-GT-202,•1994

12] J.Kurzke
Advanced User-Friendly Gas Turbine Performance Calculations on a
Figure 12: T800 compressor map Ref. I, fig 18 Personal Computer
ASME 95-0T-147, 1995
Sometimes the programs are used with calculated data as input Then it
is very easy, to draw smooth lines because there is no scatter in the (31 G.L.Converse, R.G.Giffin
calculated data. However, when the map calculation program has failed to Extended Parametric Representation of Compressor Fans and Turbines
converge in the region of pressure ratio equal to one, for example, then this Volume I
will be seen immediately. In such a case it is straightforward to ignore NASA CR-174645, 1984

20

/ 110
07
103
100
975
95
eit2 5
.

90 •

5.
c../v 80
-Lt--,\
":2
70 75
' 65
60
0
2 4 6
Corrected Airflow, lb/s

Figure 13: T800 Compressor map derived from figure 12 with SMOOTH_C