See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315649974

Computational Technique of Plotting Campbell Diagrams for Turbine Blades

Chapter · March 2017

DOI: 10.1007/978-3-319-53363-6_5

CITATIONS READS
2 659

2 authors:

Vladimir Eliseev Artem A. Moskalets

Peter the Great St.Petersburg Polytechnic University Peter the Great St.Petersburg Polytechnic University
119 PUBLICATIONS   562 CITATIONS    12 PUBLICATIONS   29 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Antisway control of gantry cranes View project

Eulerian Mechanics of Belts View project

All content following this page was uploaded by Vladimir Eliseev on 12 November 2019.

The user has requested enhancement of the downloaded file.

Computational Technique of Plotting
Campbell Diagrams for Turbine Blades

Vladimir V. Eliseev and Artem A. Moskalets

Abstract The technique of plotting the Campbell diagrams for turbine blades is
developed by means of mathematical modeling. Equations of blade oscillations in
the field of centrifugal forces are derived. The turbine blade is supposed to be a
naturally twisted rod. The numerical solution of the eigenvalue problem gives the
required dependencies of oscillation frequency on angular velocity. Calculations are
performed using the shooting method in Mathcad. The computational results for a
particular turbine blade are presented.

Keywords Turbine blade
Campbell diagram
Naturally twisted rods
ODE

system Shooting method
Natural frequencies and modes
Mathcad

Problem

Blades on rotating turbines are influenced by steam or fluid jets generating periodic
excitation with rotational speed X. A blade as an oscillating system has a set of
natural frequencies xk (in theory, its number is infinite). In order to prevent reso-
nances, the inequality X 6¼ xk must be satisfied. However, this statement should be
clarified.
Firstly, the harmonics of frequencies 2X; 3X; . . . arise from the load Fourier
expansion. Secondly, the natural frequencies depend on angular velocity, because
the increase of centrifugal forces causes an increase in bending stiffness. Therefore,
this condition holds:

nX 6¼ xk ðXÞ ð1Þ

V.V. Eliseev
A.A. Moskalets (&)

Peter the Great Saint-Petersburg Polytechnic University, Saint-Petersburg, Russia
e-mail: artem.moskalec@gmail.com
V.V. Eliseev
e-mail: yeliseyev@inbox.ru

© Springer International Publishing AG 2017 37

A.N. Evgrafov (ed.), Advances in Mechanical Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-53363-6_5
38 V.V. Eliseev and A.A. Moskalets

for all n; k having natural values. The Campbell diagram illustrates this inequality
graphically [1–3]. There are two families of functions of X: the left side of Eq. (1)
represents rays from the origin, the right side of (1) represents curves for xk ð0Þ.
These values can be determined by means of modal analysis of a fixed blade
[4–10].
It is so important to meet the condition of Eq. (1) that expensive
Campbell-machines are created [1] to find the right side of Eq. (1) experimentally.
The purpose of this effort is to find xk ðXÞ by mathematical modeling.
One-, two- or three-dimensional models can be used for blade modeling. The
number of dimensions depends on the proportions of the blade. It is naturally
twisted and has variable section. It is naturally twisted and has variable section
[11, 12]. The problem can’t be solved by means of elementary considerations, so
the nonlinear rod theory is required [13]. Modern approaches to the estimation of a
blade’s modal parameters are represented in articles [5–7]. These approaches are
based on computer mathematics [14].

Equations of Rod Theory

These equations were derived by Kirchhoff and then generalized by Cosserat. In

modern formulation [13], these equations can be written as:

Q0 þ q ¼ q€r; M0 ¼ Q  s;
ð2Þ
j ¼ A
M; s ¼ P
s0

Here Q; M are the vectors of force and moment, respectively; ð. . .Þ0 and ð. . .Þ_
denote the derivatives with respect to material coordinate s and time t, respectively;
rðs; tÞ is the radius vector of rod’s particle; s0 ¼ r00 ; s ¼ r0 are the unit tangent
vectors of rod axis before and after deformation, respectively; j is the strain vector
of bending and torsion; A is the compliance tensor, P is the rotation tensor, q is the
mass per unit length, and q is the load distributed per unit length.
The nonlinear equations [see Eq. (2)] are required for large deformation prob-
lems. Also, they are the basis for solving the problem of the superposition of small
deformations upon finite ones. In this case, each quantity has a small increment
denoted by a wave: ~r ¼ u; P ~ ¼ h  P; q~; . . .
Here, u; h are the translation vector and the vector of small angular displacement,
respectively. The varying the system (2) yields the following system:

~0 þq
Q ~ ¼ q€ ~0 ¼ Q
u; M ~  s þ Q  ðh  sÞ;
ð3Þ
~  h  MÞ; u0 ¼ h  s
h0 ¼ A
ð M
Computational Technique of Plotting Campbell Diagrams … 39

It is a linear system, but its coefficients are determined by the state condition
before the variation and must be precomputed as functions of s and t.

Blade in the Field of Centrifugal Forces

The blade is supposed to be a straight-line, naturally-twisted rod [12] stretched by

centrifugal forces. The rotating frame of reference is used to account for inertial forces
as in the effort [15]. The system of ordinary differential equations (ODE) in com-
ponents becomes, from Eq. (3) in the case of harmonic oscillations at frequency x:

Q0x þ qx ¼ x2 qux ; Q0y þ qy ¼ x2 quy ;

Mx0 ¼ Qy þ Qhx ; My0 ¼ Qx þ Qhy ;
ð4Þ
h0x ¼ Axx Mx þ Axy My ; h0y ¼ Axy Mx þ Ayy My ;
u0x ¼ hy ; u0y ¼ hx

Here, the wave (symbol of variation) is omitted. The Cartesian coordinates x; y; z

are used: axis z is directed along the rod axis, while axes x; y are in the sectional
plane (Fig. 1).
The bending compliances Axx ; Ayy ; Axy are evaluated using simple engineering
formulas. The tension force Q is determined by the formula:

ZL
QðzÞ ¼ X2 qðR1 þ nÞdn; ð5Þ
z

where L is the length of blade, and R1 is the radius of rotor. The formula Eq. (5) is
derived from the balance equation for longitudinal forces. For the blade made at a
Russian factory, the relationship Eq. (5) is shown in Fig. 2 ðX ¼ 3000 rev=min):

Fig. 1 System of coordinates

for the blade
40 V.V. Eliseev and A.A. Moskalets

Fig. 2 Tension force

The stiffness and inertial characteristics can be variable along the blade. To
account for this, the characteristics of both end sections and a few intermediate
sections are computed, and then the functional relationships are determined by
interpolation.
Eight boundary conditions must be specified for the system of Eq. (4).
Translational and angular displacements are zero at rigid support z ¼ 0; transverse
forces and bending moments are zero at free end z ¼ L:
The problems can be solved only numerically. Note that the system of Eq. (4) is
eighth-order in the case of forced vibrations (qx ; qy are nonzero).
The equation x0 ¼ 0 must be added to that system for the free vibration case (the
order of the system Eq. (4) increases by one). The additional boundary condition is
a constant value of any quantity at one of the ends. The normalizing constant should
be included at the last stage of analysis, as is accepted in the theory of vibrations
[11, 12]. When plotting the Campbell diagrams, the case of free vibrations must be
considered.

Computation in Mathcad

The system of Eq. (4) in matrix form is:

 T
Y 0 ¼ Fðs; YÞ; Y ¼ Qx Qy Mx My hx hy ux uy x2 ð6Þ

where the form of the column F is clear from Eq. (4).

The boundary value problem for system Eq. (6) is solved by means of Mathcad
using the built-in functions sbval–rkfixed [14]. Calculated values of the first five
natural frequencies are:

x1 ¼ 741; x2 ¼ 1; 59
103 ; x3 ¼ 3; 95
103 ;

x4 ¼ 7; 47
103 ; x5 ¼ 1; 28
104

The corresponding eigenmodes are shown in Fig. 3.

Computational Technique of Plotting Campbell Diagrams … 41

Fig. 3 Normalized eigenmodes

As shown in Fig. 4, these graphs are quite similar to those for the case without
centrifugal force.
42 V.V. Eliseev and A.A. Moskalets

Fig. 4 Second eigenmode.

The solid line is the model
with centrifugal forces, and
the line with points is the
model without centrifugal
forces

However, the centrifugal forces influence the lower frequencies considerably:

Xcf  X

100% ¼ ð 35:6 9:4 5:5 3:3 1:2 Þ%;
Xcf

where Xcf ; X are natural frequencies for the case that accounts for centrifugal forces
and for the case without it, respectively.
The algorithm described above is not the only way to solve the problem. Also,
the variational method based on the Lagrange equations [7] enables evaluation of
all natural frequencies and plotting the diagrams for the right-hand side of Eq. (1).
However the expression for potential energy is needed, and centrifugal force must
be taken into account in it, which goes beyond the scopes of the present paper.

Campbell Diagram

An important stage of turbine design is plotting the Campbell diagram that repre-
sents the dependence of vibration frequencies on rotor speed. This diagram enables
finding the potential resonance regions. Initially, these diagrams were plotted using
experimental data: turbine blades were tested in the so-called Campbell-machines.
The main part of this facility is the vacuum chamber which contains the observable
bladed disc. The blades are excited by steam or water jets; electromagnets may be
used also.
Nowadays, there is a possibility of plotting the Campbell diagrams using a
computational approach. In this effort, the foregoing theory and Mathcad [14] are
used. The diagram for the blade under consideration is shown in Fig. 5.
Computational Technique of Plotting Campbell Diagrams … 43

Fig. 5 Campbell diagram for

first three natural frequencies

However, the curve of natural frequencies is represented as a band, because the

frequencies have a spread of values. The permissible difference between the
maximum and minimum values of natural frequency is regulated by specifications.
The resonance region is the area defined by points of intersection between the
frequency band and the lines of harmonics. The resonances can be avoided by
detuning. In the case of transient motion, the dangerous resonant regime must be
passed through as quickly as possible [7], although the turbine can sometimes
operate in the resonance mode due to damping.
It should be noted that the Campbell diagram is intended for estimating the resonance
regimes graphically. However, the foregoing technique makes it possible to find all
these regimes simultaneously. Assuming in Eq. (4) x ¼ n X, the resonance speed
values X can be found. It appears that these diagrams become useless. However, since
the set of resonance speeds is quite dense, therefore the process of speed estimation by
the shooting method is laborious. Nevertheless, the Campbell diagrams are effective.
 44 V.V. Eliseev and A.A. Moskalets

Conclusion

The elementary computational technique of plotting the Campbell diagrams was

developed. The equations of small deformations upon the finite stress-strain state
were used as the theoretical basis. The shooting method was employed to solve the
boundary eigenvalue problem for ninth-order ODE systems with Mathcad.

Acknowledgements This work is done as the part of the state task with the financial support of the
Ministry of Education and Science of the Russian Federation (Project No. 933-2014, 1972–2014).

References

1. Kirillov ON (2009) Campbell diagrams of weakly anisotropic flexible rotors. Proc R Soc A:
Math Phys Eng Sci 465(2109):2703–2723. doi:10.1098/rspa.2009.0055
2. Lee CW, Han DJ (2008) Strength of modes in rotating machinery. J Sound Vib 313(1):
268–289. doi:10.1016/j.jsv.2007.11.038
3. Lee CW, Seo YH (2010) Enhanced Campbell diagram with the concept of H∞ in rotating
machinery: Lee diagram. J Appl Mech 77(2):021012. doi:10.1115/1.3173610
4. Ritto TG, Lopez RH, Sampaio R, Souza de Cursi JE (2011) Robust optimization of a flexible
rotor-bearing system using the Campbell diagram. Eng Optim 43(1):77–96. doi:10.1080/
03052151003759125
5. Eliseev VV, Moskalets AA, Oborin EA (2016) One-dimensional models in turbine blades.
Advances in mechanical engineering, Lecture notes in mechanical engineering, Springer,
Berlin, pp 93–104. doi:10.1007/978-3-319-29579-4
6. Eliseev VV, Moskalets AA, Oborin EA (2015) Deformation analysis of turbine blades based
on complete one-dimensional model. Russ J Heavy Mach 5:35–38 (in Russian)
7. Eliseev VV, Moskalets AA, Oborin EA (2015) Applying of Lagrange equations to calculation
of turbine blade vibration. Handbook Eng J 8:21–24. doi:10.14489/hb.2015.08 (in Russian)
8. Bloch HP, Singh MP (2009) Steam turbines: design, applications and rerating. McGraw-Hill,
New York, p 433
9. Brondsted P, Nijssen RPL (2013) Advances in wind turbine blade: design and materials.
Woodhead Publishing Limited, Cambridge, p 484, doi:10.1533/9780857097286
10. Leyzerovich AS (2005) Wet-steam turbines for nuclear power plants. PennWell, USA, p 481
11. Eliseev VV, Piskunov VA (2016) Composite gas-turbine Blades. Russ Eng Res 36(10):
819–822 (in Russian). doi:10.3103/S1068798X16100087
12. Khanh CL (1999) Vibrations of shells and rods. Springer, Berlin, p 423, doi:10.1007/978-3-
642-59911-8
13. Eliseev VV (2003) Mechanics of elastic bodies. St. Petersburg State Polytechn, University
Publishing House, St Petersburg, p 336 (in Russian)
14. Kiryanov D, Kiryanova E (2007) Computational science. Infinity Science Press, Hingham
15. Yeliseyev VV (2015) Dynamics, critical speeds and balancing of thermoelastic rotors. Lecture
Notes in control and information sciences. Advances in mechanical engineering, Springer,
Berlin, pp 129–136. doi:10.1007/978-3-319-15684-2

View publication stats