X T
F
K
dX
M
> 0 (4)
In equation (4):
T = one full period of the harmonic
(12)
It is therefore mandatory to perform a Finite Element Analysis
(or hammer ping test) of the final geometry for the IGV. It is
standard procedure to produce a Campbell diagram and make
sure there are no resonances between the aforementioned
frequencies and the IGV natural frequencies with typical safety
separation margins in the order of 10-20%.
It appears clear at this point that impeller and inlet guide vane
designs are tightly connected and engineers should ideally
proceed through step-by-step refinements and an iterative
process of analysis and verification.
As a matter-of-fact, it is more practical to start from a given
impeller Interference Diagram, select the value of K (number
of blades of the IGV) that would not cause any resonances and
proceed with the vane design.
It is also true that most manufacturers have master models of
typical geometries with a defined number of vanes; it is
anyway quite simple, through automated drawing techniques,
to obtain an updated model with the desired number of vanes.
A typical profile for an Inlet Guide Vane is from the NACA
symmetric series as the one shown in figure 3. A shroud can be
added to increase the stiffness of the first bending mode of
each blade and possibly avoid resonances.
Through parametric CAD techniques, a 3D model like the one
shown below can be easily and quickly adjusted and optimized.
Parameters could, for example, be defined to represent:
- Blade height at trailing edge
- Blade height at leading edge
- Mean chord
- Metal blade angle at leading edge
- Metal blade angle at trailing edge
- Number of vanes
- Maximum blade thickness
- Distribution of thickness along the chord
- Fillets between blade and disk
Such features would allow a quick regeneration of the model
each time one of the parameters needs changing. A direct link
to any commercial Finite Element software, finally, allows
keeping track of the natural frequencies and mode shape with
each change.
Figure 3: Typical symmetric NACA profile for IGV blade.
The following paragraph details a case study in which all the
presented concepts have been applied.
5 Copyright 2009 by ASME
The process flowchart adopted to achieve a valid IGV design is
represented below (Fig. 4).
IMPELLER GEOMETRY AND MATERIAL IS
SELECTED THROUGH THERMODYNAMIC
DESIGN / CENTRIFUGAL STRESSES
EVALUATION / ANALYSIS OF GAS
COMPOSITION
FE MODAL ANALYSIS AND DEFINITION OF
INTERFERENCE DIAGRAM WITH
COMPRESSOR SPEED RANGE, IMPELLER
NATURAL FREQUENCIES AND MODE
SHAPES
A REASONABLE NUMBER OF IGV BLADES K
IS CHOSEN NOT TO HAVE RESONANCES
WITH FIRST IMPELLER WITHIN THE
COMPRESSOR SPEED RANGE
IGV BLADE SHAPE IS OPTIMIZED TO
PROVIDE PROPER AERODYNAMIC ANGLES,
SOLIDITY AND LOW LOSSES
OPTIONAL CFD ANALYSIS OR MODEL
TESTING IS PERFORMED ON IGV
MODAL ANALYSIS IS PERFORMED ON IGV
TO MAKE SURE THERE ARE NO
RESONANCES WITH 1X,2X MECHANICAL
FREQUENCIES AND FIRST IMPELLER BLADE
PASSING FREQUENCY
O
P
T
I
M
I
Z
A
T
I
O
N
C
Y
C
L
E
IMPELLER GEOMETRY AND MATERIAL IS
SELECTED THROUGH THERMODYNAMIC
DESIGN / CENTRIFUGAL STRESSES
EVALUATION / ANALYSIS OF GAS
COMPOSITION
FE MODAL ANALYSIS AND DEFINITION OF
INTERFERENCE DIAGRAM WITH
COMPRESSOR SPEED RANGE, IMPELLER
NATURAL FREQUENCIES AND MODE
SHAPES
A REASONABLE NUMBER OF IGV BLADES K
IS CHOSEN NOT TO HAVE RESONANCES
WITH FIRST IMPELLER WITHIN THE
COMPRESSOR SPEED RANGE
IGV BLADE SHAPE IS OPTIMIZED TO
PROVIDE PROPER AERODYNAMIC ANGLES,
SOLIDITY AND LOW LOSSES
OPTIONAL CFD ANALYSIS OR MODEL
TESTING IS PERFORMED ON IGV
MODAL ANALYSIS IS PERFORMED ON IGV
TO MAKE SURE THERE ARE NO
RESONANCES WITH 1X,2X MECHANICAL
FREQUENCIES AND FIRST IMPELLER BLADE
PASSING FREQUENCY
O
P
T
I
M
I
Z
A
T
I
O
N
C
Y
C
L
E
Figure 4: Flowchart for IGV design/optimization.
The optimization cycle is performed until the combined
performance of the IGV/Impeller system, conceived as a
whole, is satisfactory from a mechanical and aerodynamic
point of view.
CASE STUDY
A brief description of a case study is presented here in
order to provide an example of application for the concepts
described in the previous sections.
Machine Description
The centrifugal compressor described in this paragraph is
a gas turbine driven pipeline compressor with a 60-105%
speed range. It has been in operation since the early 90s and is
equipped with an old design IGV set (only 4 vanes and very
low solidity). Since its installation on-site, the compressor has
shown a vibration spectrum with significant asynchronous
content (similar to the one displayed in Figure 5), whose
appearance has been immediately connected with the poor IGV
design.
Figure 5: Vibration spectrum of machine with old-design IGV.
The first impeller of the rotor has 17 blades and its
interference diagram is shown in Figure 6.
0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 2 4 6 8
Mode Shape - Nodal Diameter
F
r
e
q
u
e
n
c
y
[
H
z
]
Figure 6: Interference diagram of the first impeller of the rotor.
6 Copyright 2009 by ASME
The impeller is a large closed-type 3D one (i.e. the impeller
blades evolve from leading to trailing edge not only in the
radial section of the gas flow-path but starting from the axial
section). In order to tackle the vibration issues, it was proposed
to change the existing IGV set with another one being
developed for different applications. The parametric master
model was a 16 vanes high solidity design with tapered blades.
The geometry was adjusted to fit the inlet volute design of the
existing compressor and final drawings were generated before
performing the final resonance check with the impeller
interference diagram.
0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 2 4 6 8
Mode Shape - Nodal Diameter
F
r
e
q
u
e
n
c
y
[
H
z
]
N
B
=17 K=abs(M-i N
B
)
M 0 1 2 3 4 5 6 7 8
i=0 0 1 2 3 4 5 6 7 8
i=1 17 16 15 14 13 12 11 10 9
i=2 34 33 32 31 30 29 28 27 26
i=3 51 50 49 48 47 46 45 44 43
Figure 7: Interference diagram with exciting force.
It can be clearly seen from Figure 7 that, within the operating
speed range of the compressor, an exciting force with K=16
(such as the one generated downstream the IGV set) would
generate a resonance with the natural mode of the impeller
characterized by one nodal diameter and highlighted in the
chart. It is also of fundamental importance, once the Finite
Analysis model is produced, to interpret the different mode
shapes, in order to assess the risk of failure associated with
them. Relative amplitudes and normalized stress distribution
help in determining whether the risk associated with a
resonance is relevant or not. Such analysis, anyway, is out of
the scope of the present investigation. Based on the results
shown in Figure 7, it was decided to generate a new IGV
model with 14 vanes. Figure 8 displays the safety of the choice.
0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 2 4 6 8
Mode Shape - Nodal Diameter
F
r
e
q
u
e
n
c
y
[
H
z
]
N
B
=17 K=abs(M-i N
B
)
M 0 1 2 3 4 5 6 7 8
i=0 0 1 2 3 4 5 6 7 8
i=1 17 16 15 14 13 12 11 10 9
i=2 34 33 32 31 30 29 28 27 26
i=3 51 50 49 48 47 46 45 44 43
Figure 8: Interference diagram with new exciting force.
The model that had already been finalized (as shown in Figure
9) was discarded and a new design process was launched.
Figure 4 details such process:
- the impeller interference diagram is generated
- the number of vanes is chosen for the IGV, in order to
avoid any resonances. In this particular case it was
decided to have K=14
- a new IGV blade is optimized
- 3D CAD model is generated
- CFD analysis is performed on the IGV+Impeller
assembly to check aerodynamic performance
- 3D Finite Element model and mesh is generated from
the 3D CAD drawings
- modal analysis is performed on the IGV set to make
sure potential resonances are avoided (with blade
passing frequency and 1X, 2X mechanical
frequencies)
7 Copyright 2009 by ASME
At this point the validation of the model can be considered as
completed as all possible interactions between the vanes and
the impeller have been taken into account. Manufacturing of
the actual component can therefore be started.
Figure 9: Discarded IGV model with 16 vanes.
An example of CFD analysis on an IGV set of blades is shown
in Figure 10. Velocity contour plots are visualized to assess
incidence angles, boundary layer separation or vortex
formation.
Figure 10: CFD analysis on IGV.
The manufacturing and the assembly of the component was
followed by field testing.
The new vibration spectrum, as shown in Figure 11, has a
much lower content in terms of sub-synchronous frequencies
and was deemed as acceptable by the customer.
Figure 11: Vibration spectrum of machine with new-design
IGV.
CONCLUSION
A case study was provided as an example of application of
interference diagrams for centrifugal compressor impellers.
The starting point of the investigation was a vibration issue
during on-site operation of a pipeline compressor. The
presence of sub-synchronous vibration, visible on the spectrum
in Figure 5, particularly during high flow operating conditions,
has immediately been connected to an improper IGV design.
The aerodynamic forces acting on the portion of the shaft
upstream the first impeller are, in the right portion of the
compressor maps, of the same order of magnitude of the
compressor weight. The turbulent nature of the flow caused by
the low-solidity of the inlet vanes therefore induces low
frequency vibration.
The necessity of a re-design of the IGV set is the starting point
of the current investigation.
The interference diagram for the first impeller of the rotor is
generated from the Finite Element results. The original design
of the vanes does not create any resonances on the impeller
natural modes of vibration.
Through the application of a step-by-step design process as the
one described in the flow-chart of Figure 4, the new geometry
of the vanes takes form based on the results of aerodynamic
and mechanical calculations.
It is important to note that any interference issue connected
with the impeller downstream the IGV is excluded a-priori
based on the interference diagram.
The use of an existing geometry to replace the old one was
excluded based on this approach.
It should be noted, as a corollary, that further analyses lead to
the conclusion that other parameters should be taken into
account for evaluating the impact of inlet guide vanes on
impeller resonance; the distance along the flow-path between
8 Copyright 2009 by ASME
the vanes trailing edge and the impeller leading edge, for
example, could be enlarged so that the wakes of the IGV and
the pressure cyclic pulsations along the 360 are actually
dampened and no longer a dangerous source of excitation.
Charts are therefore being prepared to include such parameter
in the investigation. It is anyway recommended to avoid the
possibility of resonances a-priori if that is feasible; the
application of interference diagrams is simple and
straightforward enough to be considered as a very effective tool
to prevent impeller failures.
ACKNOWLEDGMENTS
I would like to thank the ex-colleagues in Italy that helped
me in putting together the material for this paper and Murari
Singh for explaining and detailing the concepts and the
applications of the SAFE Interference Diagram during a
Workshop at the GE Oil&Gas headquarters in Florence, Italy.
REFERENCES
[1] Singh, M. P. and Vargo, J. J., 1989, Reliability Evaluation
of Shrouded Blading Using the SAFE Interference Diagram,
Journal of Engineering for Gas Turbine and Power, 111, pp.
601-609.
[2] Singh, M. P. and Vargo, J. J., Schiffer, D. M., and Dello, J.
D., 1988, Safe Diagram- A Design Reliability Tool for
Turbine Blading, Proceedings of the Seventeenth
Turbomachinery Symposium, Turbomachinery Laboratory,
Texas A&M University, College Station, Texas, pp. 93-101.
[3] Singh, M. P. and Thakur, B. H., and Sullivan W. E., 2003,
Resonance Identification for Impellers, Proceedings of the
Thirty-second Turbomachinery Symposium, Turbomachinery
Laboratory, Texas A&M University, College Station, Texas pp.
59-70.