Journal of Applied Fluid Mechanics, Vol. 9, No. 5, pp. 2149-2160, 2016.

Available online at www.jafmonline.net, ISSN 1735-3572, EISSN 1735-3645.

DOI: 10.18869/acadpub.jafm.68.236.25222

Axial-Flow Compressor Performance Prediction in

Design and Off-Design Conditions through 1-D
and 3-D Modeling and Experimental Study
A. Peyvan and A. H. Benisi†
School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran

†Corresponding Author Email: Hajilouy@ sharif.edu

(Received June 28, 2015; accepted December 11, 2015)

ABSTRACT

In this study, the main objective is to develop a one dimensional model to predict design and off design
performance of an operational axial flow compressor by considering the whole gas turbine assembly. The
design and off-design performance of a single stage axial compressor are predicted through 1D and 3D
modeling. In one dimensional model the mass, momentum and energy conservation equations and ideal gas
equation of state are solved in mean line at three axial stations including rotor inlet, rotor outlet and stator
outlet. The total to total efficiency and pressure ratio are forecasted using the compressor geometry, inlet
stagnation temperature and stagnation pressure, the mass flow rate and the rotational speed of the rotor, and
the available empirical correlation predicting the losses. By changing the mass flow rate while the rotational
speed is fixed, characteristic curves of the compressor are obtained. The 3D modeling is accomplished with
CFD method to verify one dimensional code at non-running line conditions. By defining the three-
dimensional geometry of the compressor and the boundary conditions coinciding with one dimensional model
for the numerical solver, axial compressor behavior is predicted for various mass flow rates in different
rotational speeds. Experimental data are obtained from tests of the axial compressor of a gas turbine engine in
Sharif University gas turbine laboratory and consequently the running line is attained. As a result, the two
important extremities of compressor performance including surge and choking conditions are obtained
through 1D and 3D modeling. Moreover, by comparing the results of one-dimensional and three-dimensional
models with experimental results, good agreement is observed. The maximum differences of pressure ratio
and isentropic efficiency of one dimensional modeling with experimental results are 2.1 and 3.4 percent,
respectively.

Keywords: Axial compressor; One dimensional modeling; Three dimensional modeling; Characteristic curve.

NOMENCLATURE

A cross section area  density

C absolute velocity  cp/cv
Ca axial velocity  absolute angle(degree)
Cp constant pressure specific heat capacity  relative angle(degree)
c profile chord  loss coefficient
D diffusion factor  quantity difference
i incidence angle(degree)  work done factor
Ksh blade shape parameter  deviation angle(degree)
 camber angle(degree)
Kt ,i design incidence angle thickness
 solidity
correction factor  dimensionless incidence angle
Kt , design deviation angle thickness
Super script
correction factor * design condition
L stagnation temperature to static Sub script
temperature ratio 0 stagnation temperature
M Mach number
1 rotor inlet station
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

Mp mass parameter 2 rotor outlet station

m slop parameter 3 stator outlet station
m mass flow rate pro profile
N rotational speed of rotor end end-wall
P pressure clear clearance
R gas constant shock shock
r radius ps positive stall
res residual
ns negative stall
s entropy
max maximum
T temperature
t airfoil thickness eq equivalent
rel relative
uj rotor linear velocity
W relative velocity

1. INTRODUCTION compressors and succeeded to develop Lieblein’s

model and provided more accurate model to
Axial flow compressor performance prediction is estimate the blade profile loss. Another important
associated with the complexities and difficulties. source of energy loss in axial compressor is wall
Many studies have been conducted on the various boundary layer on which Koch and Smith (1976)
components of the axial compressor but a general conducted extensive studies and proposed accurate
model that can predict the performance model to predict the pressure loss. By modifying the
specifications of various types of axial flow correlation which predicts the pressure loss related
compressors with good accuracy is needed (Howell to tip clearance of centrifugal compressor, Aungier
and Calvert 1978; Wennerstrom 1990). Although, (2003) was able to predict tip clearance loss of rotor
many theoretical and experimental models have of axial flow compressor.
been developed to predict different losses of axial According to distortion of flow during the off-
compressor, including incidence and deviation design performance of the compressor, it is difficult
angles for design and off design conditions, to develop a model to predict the performance. In
incidence angles for negative and positive stall this context, Bullock and Johnson (1965) proposed
conditions, and stall onset, common existing an empirical relationship to calculate the deviation
experimental models have been obtained from angle of off-design condition. Aungier (2003)
experiments on a specific compressor and since the introduced the normalized incidence angle to
flow field dramatically changes due to a slight calculate the overall loss coefficient in off-design
change in geometry, or in component connected to condition.
the compressor or in gas turbine configuration, a
presented experimental correlation may not provide All the aforementioned models as well as the
good results for other cases. Furthermore, in the secondary flow loss of Howell (1958) are used for
most performance prediction studies, axial 1D modeling in this research.
compressor has been treated as an isolated
Madadi and Hajilouy (2008) developed one-
component. However, this study focuses on an
dimensional model to predict design and off-design
operating axial compressor of a gas turbine engine.
performance of two stage axial compressor.
In fact, a one-dimensional model, which predicts
However the case study, NASA compressor, was an
the performance of an operational compressor and
isolated compressor without considering the whole
its results are verified with experimental results, can
assembly of gas turbine system.
be used to redesign or improve the performance of
the compressor of the gas turbine system. Also, it is Song et al. (2001) provided a performance
plausible to use 1D modeling for parameter study predicting method which differs from the
because it is so fast. conventional sequential stage-stacking method in
which it employs simultaneous calculation of all
Lieblein (1965) tested two-dimensional cascade in
interstage variables like temperature, pressure and
order to calculate blade profile loss using the
flow velocity. The effect of interstage air bleeding
definition of equivalent diffusion coefficient. He
on compressor performance is also demonstrated on
could also obtain correlation to predict incidence
their work.
and deviation angles for the design point. Bullock
and Johnsen (1965) studied design incidence and Garzon and Darmofal (2003) used a probabilistic
deviation angles and they offered a model for methodology to investigate the impact of geometry
calculating these angles at design point. Herrig et on compressor performance. They applied a mean-
al. provided an approximate equation to calculate line multistage compressor model with probabilistic
the positive and negative stall angles for NACA 65 loss and turning models from the blade-passage
series blades. The equation was obtained from a analysis to quantify the blade geometry effects on
series of two-dimensional cascade test results with pressure ratio and efficiency.
low Mach number Aungier (2003). Koch and Smith
(1976) conducted extensive studies on various Spina (2002) presented a model to predict the whole

2150
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

gas turbine performance which uses compressor and radial and circumferential directions.
turbine performance maps obtained by using
generalized stage performance curves matched by 2. Constant pressure specific heat coefficient
means of the “stage–stacking” procedure. The change is neglected.
overall multistage compressor performance is 3. In design condition, inlet flow collides the
predicted using generalized relationships between blade at design incidence angle.
stage efficiency, pressure coefficient and flow
coefficient. 4. Flow viscosity doesn’t change due to small
variations of temperature.
Here, because the tests are performed on an axial
compressor within a gas turbine engine, only the 2.2 Inputs
running line could be obtained because it is Two types of inputs must be defined for one-
impossible to change the mass flow rate while the dimensional modeling: flow properties and
rotational speed of the engine is fixed. Therefore, compressor geometry. The entry requirements for
for each rotational speed there is only a single point design and off-design conditions are not the same.
to compare and validate the one-dimensional Not only mass flow rate isn’t defined as an input for
modeling results. To compare and verify the design condition but also with regard to inlet flow
modeling results for non-running line conditions, a angle, which is calculated with design condition
CFD modeling is applied using a verified correlation, design mass flow rate is calculated. On
commercial code. the other hand, for off-design, inlet mass flow is
Three-dimensional modeling is performed using defined as an input. Information required as input
computational fluid dynamics. Having three- for design and off-design conditions are presented
dimensional geometry of the compressor, it is in Table 1 and Table 2, respectively.
meshed. Overall number of grid for rotor and stator
domain is about 500,000. Then the boundary
Table 1 Flow inputs for 1D modeling
conditions are defined. In this process, turbulence
design Off-design
model and flow are considered to be k-epsilon and
Rotor inlet total Rotor inlet total
compressible, respectively. Total pressure and total
temperature temperature
temperature are specified as rotor inlet boundary
conditions and mass flow rate is specified for stator Rotor inlet total
Rotor inlet total pressure
outlet. In each rotational speeds, by solving the flow pressure
field with different mass flow rates, the Absolute velocity angle Absolute velocity
characteristic curves of the compressor are at rotor inlet angle at rotor inlet
obtained. Rotor rotational speed Rotor rotational speed
Mass flow rate
The experiments have been done for an axial
compressor embedded in a gas turbine in Gas
Turbine laboratory of Sharif University. There are Table 2 Geometric inputs for 1D modeling
several measuring stations on gas turbine engine Rotor and stator
and only the first four of them are used for this Inlet hub radius
study including bell mouth throat, axial compressor Inlet shroud radius
rotor inlet and outlet and finally axial compressor Outlet hub radius
stator outlet. At these stations, the stagnation and Outlet shroud radius
static pressures and temperatures as well as the Inlet camber angle of blade
engine speed and mass flow rate are measured. All Outlet camber angle of blade
measuring devices are calibrated prior to tests and Stagger angle
actual values are calculated using calibration Blade number
curves. Chord length
Maximum thickness to chord ratio
2. ONE-DIMENSIONAL MODELING Maximum thickness distance to chord ratio
Tip clearance
One dimensional modeling is performed to predict Blade roughness
design and off-design performance of the axial Axial gap between rotor and stator
compressor.
For one dimensional modeling, equations of To predict the performance of the compressor and
conservation of mass, momentum, and energy, as obtain its characteristic curves, it is necessary to
well as equations of ideal gas, second law of calculate compressor performances at design and
thermodynamics, the energy losses and predicted off-design points.
deviation angles in design and off-design conditions
are solved together on mean radius of compressor 2.3 Design Condition
blades in order to determine the air flow unknowns.
Figure 1 shows schematic of the axial compressor
2.1 Assumptions and some geometric parameters. Stations 1, 2 and 3
represent rotor inlet, rotor outlet and stator outlet
Main assumptions of 1D modeling are as below:
respectively.
1. Air flow properties changes are ignored in First with knowing geometric specifications of the

2151
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

compressor such as inlet and outlet angle of camber stagnation temperature and pressure at stator inlet
line and maximum thickness to chord ratio, it is are considered the same as calculated at rotor exit.
possible to compute design incidence and deviation With these assumptions it is possible to compute all
angles using Eqs. (1) and (2). parameters at stator inlet. The next step is obtaining
unknowns at stator exit by algorithm in Fig. 3.
i *  K shK t ,i (i 0* )10  m (1)
The outcome of design condition computations is to
calculate profile loss coefficients in design
  K shK t , 
*
  *
0 10
 m (2) condition for rotor and stator rows.

Given the inlet stagnation pressure and temperature, 2.4 Off-design Condition
the rotor rotational speed and angles of velocity
After determination of design profile loss
triangle at rotor inlet, design parameters can be
coefficients, it is affordable to obtain profile loss
calculated according to Eqs. (3) to (14) “Peyvan
coefficients in off design condition. The coefficient
(2013).
can be obtain through Eqs. (15) and (16). Mass flow
u1   rm 1N / 30 (3) rate is the code input for off-design condition.
Consequently, for one stage of axial compressor, all
Ca1  u1 /(tan(1 )  tan( 1 )) (4) the unknown quantities for off design condition in
the rotor inlet, rotor exit and stator exit are
C 1  Ca1 / cos(1 ) (5) calculated.
To obtain the off-design loss coefficients a
W1  Ca1 / cos( 1 ) (6)
dimensionless parameter for incidence angle is
defined as follow “Aungier (2003):
T1 T01 C 12 /2C p (7)
  (i  i * )/(i ps  i * ) i  i * (15)
T01rel T1 W /2C p
1
2
(8)
  (i  i * )/(i *  i ns ) i i* (16)
M 1  C 1 /  RT1 (9)
i ps and i ns are related to incidence angle of positive
M 1rel W1 /  RT1 (10)
and negative stall and i  is the design incidence
angle that is used in correlations represented by
P1  P01 /(1  0.5(  1)M 12 ) / 1 (11)
Herig Aungier (2003). Herig presented approximate
relationships to calculate positive and negative stall
 1  P1 / RT 1 (12) incidence angle for NACA 65 series.
P01rel  P1(T01rel /T1 ) / 1 (13)   30 
0.48
  (17)
i *  i ns  9  1   


  1ns   4.176

m  1 A1Ca1 (14)
 1 ps   (18)
i ps  i *  10.3  2.92  
 15.6  8.2

Finally off-design loss coefficients are obtained, as

follows:

  shock   *[1   2 ] 2    1 (19)

  shock   *[5  4(  2)]   2 (20)

  shock   *[2  2(  1)]   1 (21)

First, calculation is performed at the rotor inlet

station, based on Fig. 4 algorithm. Thereafter; other
unknowns are determined through Eqs. (22) to (27):
Fig. 1. Schematic of meridional view of u 1   rm 1N off / 30 (22)
compressor geometry with three measurement
stations. 1  arctan((u1  C 1 sin( 1 ))/ C 1 cos( 1 ))) (23)
Therefore, the mass flow rate at the inlet of the rotor (24)
W1  C 1 cos(1 )/ cos( 1 )
for design condition is obtained and calculated
parameter at rotor inlet set as input to rotor outlet (25)
M 1rel W1 /  RT1
algorithm which is shown in Fig. 2.
After obtaining the unknown parameters at the rotor T01rel T1 W12 /2C p (26)
exit, it is assumed that the flow approaches the
stator with design incidence angle. Velocity,

2152
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

Fig. 2. Flow chart for parameters calculation at rotor outlet.

Fig. 3. Flow chart for parameters calculation at stator outlet.

P01rel  P1(T01rel /T1 ) / 1 stator exit stations with a slight change using
(27) Johnson and Bullak (1965) model to calculate
The algorithm used for calculating design condition outflow relative angle. Incidence angle and axial
is also applied for off design condition at rotor and velocity ratio of the stage are the most significant
parameters for calculation of deviation angle

2153
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

Aungier (2003). Johnson and Bullak (1965) performance of a compressor with a specific
presented a model to calculate variation of deviation geometry.
angle at design point. After a lot of experiments and
corrections they succeeded to obtain a correlation
which predicts linear variation of deviation angle
based on incidence angle, Eq. (28). According to
linear variation of deviation angle with incidence
angle and the correction that was made by Pullard
and Gastello, considering the effect of axial velocity
variation, the deviation angle for off design
condition is calculated with Eq. (29).
*
  
   1  (  0.25 )( 1 /53)  / exp(3.1 ) (28)
4 2.5

 i 

*
  
 *    (i  i )  10(1 Wa 2 /Wa 1 )
*
(29)
 i 

3. STALL AND SURGE

Despite the complexity of the surge phenomeana

there are methods that can be used to determine the
surge criteria. Aungier (2003) defined a parameter
named equal velocity ratio:

P02rel  P2
W RE  (30)
P01rel  P1

Based on Renao investigations on regular two

dimensional diffusers, Aungier (2003) determined
stall criteria according to the following equation. Fig. 4 Flow chart for parameters calculation at
rotor inlet for off-design condition.
W RE 
0.15  11t max  
/ c / 0.25  10t max / c 
(31)
 
0.65
1  0.4  / 2sin( / 2)cos   4.1 Solving Problem with Three-
Dimensional Method
Comparing the results of Eq. (31) with the
experimental results led to the relationship suitable In this research, flow field of a single stage axial
for diffusion coefficient over 2.2 as Eq. (32) compressor of the gas turbine in the laboratory is
(Aungier 2003). solved by 3D modeling. In three dimensional
modeling the equations of conservation of mass,

 2.2/ D
  0.15  11t   
0.6

 eq max
/ c / 0.25  10t max / c momentum and energy coupled with ideal gas
WRE  (32) equation and k-epsilon turbulence model are solved
 
0.65
1  0.4  / 2sin( / 2)cos  
simultaneously to achieve flow field solution Nili-
Ahmadabadi et al. (2014).
Eq. (32) is valid if relation (33) is satisfied.
(33)
4.2 Grid Construction
 / 2sin( /2)cos    1.1
Because of symmetry of flow field it is sufficient to
just take a rotor blade and a stator blade into
4. THREE DIMENSIONAL MODELING account with definition of proper boundary
condition which results in dramatically reducing
Application of numerical methods for solving time. Fig. 5 shows the flow field around a
turbomachinery simulation has started since 1940. rotor blade with boundary conditions.
In fact, these methods have been formulated before Flow fields in the rotor and stator are meshed using
digital computers advent. First, streamline curvature structured grid (Figs. 6 to 8). The topology of mesh
was presented by Wu which can be considered as a is H/J/C/L. The ratio of number of nodes of rotor
primary numerical method for turbomachinery and domain to stator domain is 60 to 40. The grid
its concept has been used wildly till 1980 when independency was performed based on the results
CFD methods were practically applicable Denton which resulted in 500,000 nodes are enough for
and Dawes (1998). rotor and stator domains overall in order to obtain
an acceptable result.
Full three dimensional CFD method considers
geometric details such as sweep angle, lean angle 4.3 Problem Setup
and wall profile of axial compressor which
results in detailed investigation of flow field. In Flow field includes various zones such as hub,
this study CFD method was used to predict the shroud, and blade walls, flow inlet boundary, flow

2154
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

outlet boundary, periodic boundaries 1 and 2, and 5. EXPERIMENTAL STUDY

interface at tip clearance. The air flow passing
through the domain is considered as an ideal gas. At Sharif University Gas Turbine Laboratory, a
The energy turbulence model is chosen to be k- small gas turbine engine, equipped in an
epsilon. Types of boundary conditions are very experimental test bed, see Fig. 9. The gas turbine
important to complete the problem definition. At engine includes a single stage axial compressor
inlet (station 1) total temperature and total pressure which is investigated in this study.
and flow direction are defined. At stator outlet the
amount of output mass flow rate is specified. Three
types of interfaces are defined: a) the interface
between rotor domain outflow surface and stator
inflow surface, b) periodic interface at both sides of
rotor and stator domain, and finally, c) the interface
at rotor tip clearance which connects pressure side
and suction side of the blade tip, see Fig. 8.
Moreover, the rotational speed of the rotor is
defined.

Fig. 7. Rotor blade and the middle control layer.

Fig. 8. Rotor and stator blades, left and right

sides respectively.
Fig. 5. Flow field and boundaries.

Fig. 9. Test setup.

5.1 Measurement and Calibration System

There are twelve measuring stations installed on the
engine in axial direction, those related to this
research are as follows:
1. Inlet of bell mouth, 2. Rotor inlet of axial
compressor, 3. Rotor exit, 4. Exit of axial
compressor.

5.1.1 Measurement System

Fig. 6. Mean surface of rotor (upper) and stator Each of quantities are measured by special
(lower) grid. equipment as following:

2155
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

A. Pressures: all static pressures are measured using enthalpy of the compressor increases which causes
four static pressure taps equally spaced around the more pressure augmentation. In Fig. 12, the
circumference of the duct, and joined together and compressor isentropic efficiency is plotted versus
connected to a strain gauge pressure transducer, to the mass parameter.
give an average static pressure. The total pressures
are measured using Pitot tubes connected to the
strain gauge pressure transducers.
B. Temperatures: Temperatures are measured using
J type thermocouples. The maximum error of the
temperature measurement system is about 0.25 0C
Pourfarzaneh (2010). All thermocouples are
connected to the computer data acquisition system
by Adams cards Asgarshamsi et al. (2014).
C. Air mass flow rate: The compressor air mass
flow rate is measured using a bell mouth entry flow
meter. The mass flow calculation is based on Fig. 10. Stagnation Pressure ratio versus mass
pressure drop measurement across the bell mouth
parameter Mp  m (kg / s ) T0 (k ) / P0 (bar )
throat and ambient temperature and pressure
measurements. (experimental results).
D. Rotational speed: The compressor rotational
speed is measured using a magnetic pick-up, which
counts shaft revolution per second.
5.1.2 Calibration
For calibration of the pressure gauges 2 meters long
manometer filled with mercury is used. It is
accomplished by measuring different values of
pressures by data acquisition system as well as
manometers. Thereafter, by a curve fitting the test
points with minimum error, the calibration curve is
obtained Pourfarzaneh (2010).
In order to calibrate J type thermocouples, some
equilibrium temperatures such as ice-water mixture Fig. 11. Stagnation Pressure ratio versus speed
temperature are used. An accurate RTD parameter N  N (rpm) / T0 (k ) (experimental
thermometer is used as well.
results).
5.2 Test Method
The compressed air is blown into the turbine blades
and consequently engine speed rises to ignition
speed. Then the fuel is injected and the ignition
system ignites the mixture of air-fuel and finally.
Gradually engine accelerates to higher speeds,
which at several steady state conditions, the
Automated Data acquisition system records all data
every second. All the data obtained from the sensors
are calibrated with calibration curves achieved from
regression of calibration data points.

5.3 Experimental Results Fig. 12. Isentropic efficiency versus mass

parameter (experimental results).
At this section the axial compressor performances
are presented. Fig. 10 shows the running line of
pressure ratio versus mass parameter, of the axial 5.4 Errors and Uncertainty of Experimental
compressor from start condition to nominal speed of Results
engine. The trend of the curve is acceptable in
To calculate the overall uncertainty caused by the
comparison with typical one Mattingly and von
combined effects of uncertainties in various
Ohain (1996). According to this curve, pressure
variables, the following general equation is
ratio increases as mass parameter and speed.
introduced Nakra and Chaudhry (2004).
In Fig. 11, the compressor pressure ratio changes
y  f (x 1 , x 2 ,..., x i ,..., x n ) (34)
versus speed parameter are shown. In this figure
speed parameter increases from idle to rated speed,
while pressure ratio the increases from 1.05 to The parameter y depends on independent
1.402. With increasing rotational speed, linear variables x1 ، x2 ،...، xi ،...، xn . Uncertainty caused by
speed of the blade increases, therefore output

2156
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

Table 3 Uncertainty values for three values of mass parameter

M P
 m  C T L Mass
parameter
0.380822 0.019618 0.003742 0.003483 1.000301 0.000471 104.37
0.436719 0.0022498 0.003728 0.00228 0.997325 0.000475 119.98
0.525062 0.027049 0.003701 0.00128 0.991638 0.000438 142.51

the combined effects of different variables can be a given rotational speed the curve of pressure ratio
written as: versus mass parameter has a negative slope. The
curve to the left side and right side is limiting to
1/2
 n  y 2  choking and surge conditions respectively. With a
 
2
 I        xi  (35) slight increase in mass parameter near choking
 i 1  x i   condition, pressure drop increases sharply resulting
 
the declination of pressure ratio. It is observed that
The term  I is general uncertainty for y parameter the compressor maximum pressure ratio near surge
condition for design rotational speed is 1.432. Surge
and the term  xi is uncertainty relating to measured line is ascending which is the same as those are in
quantity x i . the literature.
In order to investigate flow angles, the compressor
For example, the stagnation pressure ratio blade characteristics are briefly presented in Table
uncertainty is obtained by Eq. (36). 5.
P02
 P    P 
2 2
Pr    Pr  Pr / P02 / P01 (36)
P01 02 01 Table 5 Compressor blade angles
Inlet Outlet Stagger
Experimental results which are extracted from the camber camber angle
data that is captured from three measuring stations angle angle
of gas turbine engine are used for comparison with Rotor 51.28o 23.49o 39.9o
theoretical results and uncertainty analysis is Stator 38.28o 0o 12.06o
performed on the data captured at these stations.
The values of uncertainties for mass parameter, There are several performance points which were
total pressure ratio and isentropic efficiency are modeled through 1D modeling so that it is
shown in Table 3. According to the obtained values impossible to investigate their velocity triangles.
of the uncertainty it can be concluded that the However, the velocity triangles are delineated for
sensors are suitable to be used in the test. As an three mass flow rate amount at design speed which
example for mass parameter of 104.37, are shown in Fig. 16.
uncertainties of mass parameter, pressure ratio, and
isentropic efficiency are 0.4, 0.15 and 3 percent
respectively.

Table 4 Uncertainty values of efficiency and

pressure ratio for values of mass parameter
 Mass
 Tr  Pr
parameter
0.04635 0.004952 0.001737 104.37
0.04336 0.00496 0.001798 119.98
0.033206 0.005162 0.001866 142.51

Fig. 13. Single stage of axial compressor

6. ONE DIMENSIONAL MODELING characteristic curve of pressure ratio versus
RESULTS mass parameter through 1 D modeling.

Axial compressor maps are shown in Figs 13 and According to Fig. 14 at a specific mass parameter,
14. In these diagrams stagnation pressure ratio and isentropic efficiency increases with rotational
efficiency changes versus mass parameter are speed. Similar to pressure ratio graph, efficiency
plotted in four different rotational speeds. Also curves are limiting to choking condition with
surge line of the compressor is plotted. Stall minimum magnitude at right side and surge
condition is defined according to Eqs. (31) and (32) condition at left side.
for rotor and stator rows. If parameter WRE for rotor Fig. 15 shows the ratio of required power to design
row or stator row is less than the value defined by point power of compressor versus mass parameter
Eq. (32), stall condition occurs. As the rotational for different rotational speeds. Maximum required
speed increases, more pressure ratio is obtained. For power is related to design point rotational speed at

2157
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

mass parameter near the surge condition. residuals are changed oscillatory and the
convergence rate becomes lower. Away from the
surge condition and at midpoints, convergence rate
increases and residuals do not oscillate. Near
choking condition, if mass flow rate wouldn’t be
chosen in the proper range, then solution diverges.

8. COMPARISON OF 1-D AND 3-D

MODELING RESULTS WITH
EXPERIMENTAL RESULTS

Experimental results are averaged on running line of

the gas turbine engine resulting a single point for every
Fig. 14. Single stage of axial compressor rotational speed. Therefore, from experimental results
characteristic curve of isentropic efficiency all main parameters including, rotational speed, mass
through 1 D modeling. flow rate, pressure ratio and efficiency are obtained.
According to the experimental data in Figs 10 to 12,
due to slight changes in engine condition during the
test, and massive data captured, there is some
scattering in recorded data.  Therefore, data for each
point are averaged.
The solid squares shown in Figs 19 and 20 represent
experimental results for each speed. It can be seen
that the modeling results are in good agreement
with experimental results. By connecting these
points the compressor running line in gas turbine
engine is obtained. Similarly, the efficiency lines
are shown in Fig. 20. Maximum differences of 1D
results from experimental results are at 66 percent
Fig. 15. Single stage of axial compressor of nominal rotational speed which is 1.2 and 3.8
characteristic curve of required power through 1 percent for pressure ratio and efficiency
D modeling. respectively. In Fig. 21, three-dimensional
modeling results are compared with the results of
7. THREE DIMENSIONAL MODELING one-dimensional modeling in two rotational speeds,
RESULTS which shows very good agreement. Moreover,
value of surge mass flow rate obtained by 1D
modeling can be used to estimate the more accurate
Numerical solution results change considerably
value which can be obtained with 3D modeling in
with grid size up to a certain amount. Therefore, it
which the surge condition can be specified by the
is necessary to increase the number of grid to a
numerical residuals become oscillating Mikhailova
point which the results don’t change. For 80 percent
et al. (2014). In this sense, the predicted mass
of design speed and 125.05 mass parameter, flow
parameter for surge condition through 1D and 3D
field is solved with different grid numbers. The
modeling show a slight difference which is about 10
compressor pressure ratio and efficiency results are
percent of mass parameter predicted by 1D
used for grid independency check. It is shown that
modeling at 80% design rotational speed.
by increasing the number of grids from 500,000 to
1000,000, pressure ratio and efficiency change
about 0.009 and 0.03 percent respectively. 9. CONCLUSION
Therefore, in order to reduce the computational
time with acceptable accuracy, the number of grid In this study, two methods are used to predict axial
points is chosen to be 500,000. compressor performance curves: one and three
dimensional modeling.
To obtain the characteristic curves of the
compressor using three-dimensional method, three In order to obtain maps of a particular compressor it
rotational speeds the same as one dimensional is needed to determine some characteristics
modeling are selected and flow field is solved for including the value of surge mass flow rates at
various mass flow rates ranging from surge to different rotational speeds, choking condition mass
choking conditions. flow rate, and the points in between these two
extremities. Meanwhile, by comparing the results of
Figures 17 and 18 have similar trend to one 1D and 3D, it can be concluded that the model of
dimensional modeling results. The total number of predicting stall can predict the surge condition with
grid is the same for all rotational speeds. For lower an acceptable accuracy. The value of surge mass
rotational speeds the accuracy of solution increases. flow rate obtained by 1D modeling can be used in
For a fixed geometry, with increase of speed, time order to estimate the more accurate value which can
scale reduces so for acceptable error, element size be obtained with 3D modeling in which the surge
has to be reduced Löhner (2008). In a particular condition can be specified by the numerical
speed, for mass flow rates close to surge condition,

2158
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

Mass parameter=141.8 Mass parameter=147.5 Mass parameter=155.35

Fig. 16. Velocity triangles for design rotor speed for three different mass flow rate.

residuals when they become oscillating. Moreover, with 1D modeling is that the compressor works near
similarly the maximum mass flow which is possible its surge as the engine reaches its nominal speed.
to pass through the compressor for a specific speed
is obtainable through 1D modeling with high slop Comparison between one-dimensional modeling
of efficiency and pressure ratio curves. results with experimental results shows good
Furthermore, with estimated value of choking mass agreements for different rotational speeds.
flow rate by 1D modeling the mass flow capacity of Maximum performance prediction difference for
the compressor is determined by 3D modeling so pressure ratio and isentropic efficiency are 2.1 and
that mass flow rate for a particular speed can be 3.4 percent respectively. Due to the fact that the
increased to the point above of which the solution experimental results are obtained for a compressor,
diverges. which is effected by other components of the
engine, working inside a gas turbine engine, it can
be concluded that the 1D modeling sufficiently
predict the performance of a compressor while
working in an assembly.

Fig. 17. Single stage of axial compressor

characteristic curve of pressure ratio through 3-
D modeling.
Fig. 19. Comparison of 1-D pressure ratio results
with experimental results.

Fig. 18. Single stage of axial compressor

characteristic curve of isentropic efficiency
through 3-D modeling.

Fig. 20. Comparison of 1-D isentropic efficiency

One significant finding which can be concluded results with experimental results.
from the comparison of experimental pressure ratio

2159
A. Peyvan and A. H. Benisi / JAFM, Vol. 9, No. 5, pp. 2149-2160, 2016.

dimensional cascades. NASA Special

Publication 36, 183.
Löhner, R. (2008). Applied computational fluid
dynamics techniques: an introduction based
on finite element methods. John Wiley & Sons.
Madadi, A. and A. H. Benisi (2008). Performance
Predicting Modeling of Axial-Flow
Compressor at Design and Off-Design
Conditions. ASME Turbo Expo 2008: Power
for Land, Sea, and Air 317-324.
Mattingly, J. D. and H. von Ohain (1996). Elements
Fig. 21. Comparison of 1-D pressure ratio results of gas turbine propulsion. McGraw-Hill New
with 3-D modeling results. York.
Mikhailova, A., D. Akhmedzyanov, Y. M.
REFERENCES Akhmetov and A. Mikhailov (2014). Joint
prediction of aircraft gas turbine engine axial
Asgarshamsi, A., A. H. Benisi, A. Assempour and flow compressor off-design performance and
H. Pourfarzaneh (2014). Multi-objective surge line based on the expanded method of
optimization of lean and sweep angles for generalized functions. Russian Aeronautics (Iz
stator and rotor blades of an axial turbine. VUZ) 57, 291-296.
Proceedings of the Institution of Mechanical Nakra, B. and K. Chaudhry (2004).
Engineers, Part G: Journal of Aerospace Instrumentation, measurement and analysis.
Engineering, p. 0954410014541080. Tata McGraw-Hill Education.
Aungier, R. (2003). Axial-Flow Compressors-A Nili-Ahmadabadi, M., M. Durali and A. Hajilouy
Strategy for Aerodynamic Design and (2014). A Novel Aerodynamic Design Method
Analysis. , ed: New York: ASME Press. for Centrifugal Compressor Impeller. Journal
Bullock, R. and I. Johnsen (1965). Aerodynamic of Applied Fluid Mechanics 7.
Design of Axial Flow Compressors. NASA Peyvan, A. (2013). Two-stage axial compressor
SP-36. design software. MSc Theses, Department of
Denton, J. and W. Dawes (1998). Computational Mechanical Engineering, Sharif University of
fluid dynamics for turbomachinery design. Technology, Tehran.
Proceedings of the Institution of Mechanical Pourfarzaneh, H. (2010). Turbojet engine
Engineers, Part C: Journal of Mechanical performance modeling and evaluation with
Engineering Science 213, 107-124. experimental test. PHD theses, Department of
Garzon, V. E. and D. L. Darmofal (2003). Impact of Mechanical Engineering, Sharif University of
geometric variability on axial compressor Technology, Tehran.
performance, in ASME Turbo Expo 2003, Song, T., T. Kim, J. Kim and S. Ro (2001).
collocated with the 2003 International Joint Performance prediction of axial flow
Power Generation Conference, 1199-1213. compressors using stage characteristics and
Horlock, J. H. (1958). Axial Flow Compressors. simultaneous calculation of interstage
Florida: Robert E. Krieger publishing CO. parameters. Proceedings of the Institution of
Mechanical Engineers, Part A: Journal of
Howell, A. and W. Calvert (1978). A new stage Power and Energy 215, 89-98.
stacking technique for axial-flow compressor
performance prediction. Journal of Spina, P. (2002). Gas Turbine performance
Engineering for Gas Turbines and Power 100, prediction by using generalized performance
698-703, curves of compressor and turbine stages.
ASME Turbo Expo 2002: Power for Land,
Koch, C. and L. Smith (1976). Loss sources and Sea, and Air, 1073-1082.
magnitudes in axial-flow compressors, Journal
of Engineering for Gas Turbines and Power Wennerstrom, A. (1990). Highly loaded axial flow
98, 411-424. compressors: history and current
developments. Journal of Turbomachinery
Lieblein, S. (1965). Experimental flow in two- 112, 567-578.

2160