Sie sind auf Seite 1von 19
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1355-2511.htm

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1355-2511.htm

METHODOLOGY AND THEORY

Blades condition monitoring using shaft torsional vibration signals

B.O. Al-Bedoor

Mechanical Engineering Department, Faculty of Engineering and Technology, University of Jordan, Amman, Jordan, and

S. Aedwesi and Y. Al-Nassar

King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia

Abstract

Purpose – The purpose of this paper is to validate mathematically the feasibility of extracting the rotating blades vibration condition from the shaft torsional vibration measurement. Design/methodology/approach – A mathematical model is developed and simulated for extracting rotating blades vibration signatures from the shaft torsional vibration signals. The model simulates n-blades attached to a rigid disk at setting angles and the shaft drives the disk is flexible in torsion. The model is developed using the multi-body dynamics approach in conjunction with the Lagrangian dynamics. A three-blade rotor system example is simulated for blades free and forced vibration under stationary and rotating conditions. Frequency spectrums for the shaft torsional and blades bending vibration are represented and studied for analysis verification purposes. Findings – The torsional vibration frequency spectrums showed blades free and forced vibration signatures. The blade setting angle is shown to reduce the sensitivity of torsional vibration signal to blades vibration signatures as it increases. The torsional vibration signals captured the variation in blades properties and produced broadband frequency components for mistuned system. The shaft torsional rigidity is shown to reduce the sensitivity of torsional vibration signal to blades vibration if increased to extremely high values (approaching rigid shaft). The rotor inertia is shown to have less effect on the torsional vibration signals sensitivity. The method of torsional vibration as a tool for rotating blades vibration measurement, based on the proposed mathematical model and its simulation, is feasible. Practical implications – There is a growing need for reliable predictive maintenance programs that in turn requires continuous development in methods for machinery health monitoring through vibration data collection and analysis. Turbo machinery and bladed assemblies like fans, marine propellers and wind turbine systems usually suffer from the problem of blades high vibration that is difficult to measure. The proposed new method for blades vibration measurement depends on the shaft torsional vibration signals and can be used also for verifying the signals from other types of bearings sensors for possible blades vibration condition monitoring. Originality/value – This paper presents a unique mathematical model and simulation results for the rotating blades vibration monitoring. The developed model can be simulated for studying coupled blades vibration problems in the design stage as well as for condition monitoring in maintenance applications.

Keywords Condition monitoring, Vibration, Maintenance

Paper type Research paper

Blades condition monitoring

275

Paper type Research paper Blades condition monitoring 275 The authors acknowledge the support of King Fahd

The authors acknowledge the support of King Fahd University of Petroleum. This work is funded by KFUPM Research Office under project number ME/BLADE-VIBRATION/215. The first author acknowledges the support of the University of Jordan. B.O. Al-Bedoor is on leave from King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia.

Journal of Quality in Maintenance Engineering Vol. 12 No. 3, 2006 pp. 275-293 q Emerald Group Publishing Limited

1355-2511

DOI 10.1108/13552510610685110

JQME

12,3

276

1. Introduction Rotating bladed mechanical systems are widely used as basic parts in machinery. Bladed rotors can range from systems with hundred of blades as in turbomachinery to systems with three to four blades as in fans, marine propellers and wind turbines. Regardless of the number of rotating blades difficulties were faced in measuring the rotating blades vibration due to the fact they are rotating and interacting with the working environment. Recent laboratory experiments on rotors with limited number of blades and field measurements on fans (Maynard and Trethewey, 1999, 2001; Maynard et al., 2000) showed that rotating blade vibration frequencies could be extracted from the shaft torsional vibration signal. However, due to the experimental nature of these studies and the importance of the problem of rotating blades vibration measurement, a mathematical model that enables verifying the approach and opens the road for more quantified studies in this direction is extremely needed. Vibration measurement is known as a powerful tool in machinery condition monitoring, Laws and Muszynska (1987) and Vance (1988), that put growing demands on developing reliable vibration measuring systems. These systems are required to sense and represent closely the machinery particular component vibration with minimum interference from other vibration sources. When blades vibration information is required, the task becomes very complicated, Simmons and Smalley (1990), as blades are rotating and interacting with the working environment. To directly monitor blade vibration, strain gages were used in many laboratory-testing studies (Kaufman and Kershisnik, 1984; Srinivasan and Cuts, 1984; Scalzo et al. , 1986; Fan et al. , 1994) that showed the practical technique limitations in terms of sensor survival due to the harsh operational conditions. Other techniques for blades vibration measurement were proposed; among these techniques is the use of Laser-Doppler and optical methods (Cookson and Bandyopadhyay, 1980; Nava et al. , 1994; Reihardt et al., 1995), with some problems and limitations on the speed of rotation. Detailed discussion of the available methods for blade vibration measurement is reported by Al-Bedoor (2002). Wherein, he classified methods for blades vibration measurement into two categories: the direct and the indirect methods. The approach of extracting blades vibration frequencies from the shaft torsional vibration was investigated experimentally by Maynard and Trethewey (1999, 2001) and Maynard et al. (2000). They explained the superiority of this approach to the lateral vibration approach, mainly, as torsional vibration is less affected by the boundary conditions than lateral vibration. In addition, Muszynska et al. (1992) used the torsional vibration to identify rotor crack. They came up with an explanation for the sensitivity of torsional vibration that damping in torsional vibration modes is minimal. Al-Bedoor (1999) proposed a mathematical model for the coupled blade bending and shaft torsional deformations in rotors using the finite element method. In addition, Al-Bedoor and Al-Nassar (2001) used the ANSYS finite element package to model the coupled multi-bladed disk-shaft flexible system. The ANSYS model was able to simulate the system rotating condition for natural frequencies and mode shape analysis. The package accounts for the effect of rotation in the geometrical stiffness matrix and no coupling between the rigid body and flexible modes was

affordable. Due to the dimensionality of the finite element model of reference (Al-Bedoor, 1999) and the limitation of the ANSYS model (Al-Bedoor and Al-Nassar, 2001) to account for the coupled dynamics, Al-Bedoor (2001) developed a reduced order model for the coupled shaft torsional and blade bending vibrations using the assumed modes method. This model simulated the shaft-disk-blade system driven by a torque that enabled monitoring the system vibration in its start-up. The model simulations showed strong dynamic coupling between blade bending and shaft torsional vibrations. However, the model was not oriented to account for steady state running condition and no provision was given to account for the blade setting angle and multi-bladed systems. In this study a mathematical model for a shaft-disk-blades system that is running at constant rotating speed is developed and simulated. The shaft is considered to be flexible in torsion and the disk is assumed to be rigid. The blades are modeled as attached to the rigid disk with a setting angle that allows monitoring the tangential (lead-lag) and axial (flapping) blades vibration modes. Moreover, the model allows simulating the effect of differences in blades properties that is known as the mistuning effects. The model is developed using the Lagrangian dynamics in conjunction with the assumed modes method to descritize blade deformation. The model is simulated for various transient and steady state blades vibrations and the shaft torsional vibration is monitored in time and frequency domains.

2. Model development The schematic diagram of the motor-shaft-disk-blades system is shown in Figure 1. The model is developed having the following assumptions in mind:

The system is rotating at constant running speed v that represents steady state running condition.

.

.

.

.

The disk is rigid with radius R d , mass m d and mass moment of inertia J d .

The shaft is flexible in torsion and no lateral shaft deflections are considered.

The blades are simple uniform inextensible beams and the Euler-Bernoulli beam theory is adopted. The assumed modes method (AMM) and the cantilevered

adopted. The assumed modes method (AMM) and the cantilevered Blades condition monitoring 277 Figure 1. Shaft-disk-blade

Blades condition monitoring

277

Figure 1. Shaft-disk-blade unit driven by electrical motor through flexible coupling

JQME

12,3

 

.

.

278

modes shapes are used. The blades are attached radially to the rigid disk with setting angle b i for each blade.

The shaft torsional and blade bending deformations are small.

The effect of axial dynamics known as blade stiffening is accounted for using the added potential energy that results from the inertial forces and blades

shortening.

The coordinate system used in developing the model is shown in Figure 2 that shows a model with three blades and the typical deflected configuration of one of the blades. The used axes are the inertial reference frame XYZ and the blade body coordinate system x b y b z b .

2.1. Kinetic energy expressions The kinetic energy of the shaft-disk-blades unit is constituted of the motor kinetic energy U m , the disk kinetic energy U d and the blades kinetic energy U b . The shaft inertia is lumped into the disk and motor inertia. The motor rotor and the disk are modeled as rigid inertia with mass moment of inertia J m and J d about the axis of rotation (Z ), respectively. As the system is rotating at a constant angular velocity v, the motor and disk kinetic energy expressions can be written, respectively, as:

Figure 2. Blades coordinate systems and typical blade-deflected configuration

U

m ¼

U d ¼

1 2 J m v 2

1

2

J d

_

2

v þ c

ð 1Þ

and typical blade-deflected configuration U m ¼ U d ¼ 1 2 J m v 2

where c is the small shaft torsional deformation angle measured with respect to the motor shaft. To develop the kinetic energy expression for the blades, the deformed configuration of the shaft-disk-blade(s) system, shown in Figure 2, is used. The global position vector of a material point P , on a typical blade, can written as:

R P ¼

(

R d cos u

R d sin u

) þ

b

A ð uÞ r P

ð 2Þ

where the angle u ¼ vt þ c describes the rotation of the blade coordinate system x b y b z b with respect to the inertial reference frame XYZ due to the system rigid body rotation at constant speed v and the shaft torsional deformation angle c. r P is the position vector of a material point on the deformed blade principal axis and can be written as:

b

r

b

P ¼

8

>

>

<

>

>

:

x

uð x ; t Þ cos b

uð x ; t Þ sin b

9

>

>

=

>

>

;

ð

3Þ

Blades condition monitoring

279

where uð x ; t Þ is the blade bending deflection and b is the blade setting angle. The rotational transformation matrix A ð uÞ can be written in the following form, based on the assumption of small torsional deformation:

A ð uÞ ¼

2

6

6

4

cos vt

sin vt þ c cos vt

0

2 c sin vt 2c cos vt

0

0

01

2 sin vt

2 c sin vt þ cos vt

3

7

7

5

ð

4Þ

The velocity vector of the material point in the inertial reference frame can be obtained by differentiating equation (2) that after manipulation can be represented as follows:

where:

_

R P ¼

8

>

>

<

>

>

:

2 L 1 sin vt 2 L 2 cos vt

2 L 2 sin vt þ L 1 cos vt

u_ sin b

9

>

>

=

>

>

;

L 1 ¼

L 2 ¼ R d cv

v R d þ x 2 cu cos b þ cð x þ R Þ þ u_ cos b

_

d

_

þ v x c þ u cos b þ cu cos b þ cu_ cos b

ð

ð

5Þ

6Þ

The kinetic energy of blade i can be found by using the velocity vector of equation (5) into the integral:

JQME

12,3

U bi ¼

1

2 Z L

0

r

_

R

T

P

_

R P dx

ð 7Þ

where r is the i th blade mass per unit length and L is its length. Now the total kinetic energy expression of the system can be written as follows:

280

n

U ¼ U m þ U d þ X U bi

i ¼1

where n is the number of blades.

ð 8Þ

2.2. Potential energy expressions

The system potential energy is constituted of the blades bending strain energy, V b , the

torsional strain energy, V T and the potential energy of the axial shortening due to transverse deformations and the motion generated inertial forces, V A . The i th blade-bending elastic strain energy that has flexural rigidity EI ðx Þ is given by:

V bi ¼

1

2 Z L

0

EI ð x Þ

2 u

x 2

dx;

ð 9Þ

the torsional elastic potential energy stored in the flexible coupling is given by:

V T ¼

1

2

k T c 2

ð10 Þ

The axial shortening due to transverse deformations in conjunction with the radial inertial forces contributes to the system elastic potential energy by the known (Al-Bedoor, 2001), axial shortening potential energy. The system potential energy is:

V

n

¼ V T þ X ðÞ

V

bi

þ

V Ai

i ¼1

where n is the number of blades.

ð11 Þ

2.3. Equations of motion

Now using the discretized form of the system kinetic and potential energy expressions into the Lagrange’s equation, performi ng the required differentiation and mathematical manipulation for the system degrees of freedom ( c, f q g i , i ¼ 1 ! n ), n is the number of blades, the system equations of motion can be represented in the following compact matrix form:

2

6

6

6

6

6

6

6

6

6

6

6

4

m cc

2

6

6

6

6

6

6

þ 6

6

6

6

6

4

k cc

m cq 1

m q 1 q 1

Symm

0

k q 1 q 1

Symm

m cq 2

0

m q 2 q 2

0

0

k q 2 q 2

···

···

···

.

.

.

···

···

···

.

.

.

m cq n

0

0

0

m q n q n

0

0

0

0

k q n q n

3

7

7

7

7

7

7

7

7

7

7

7

5

3

7

7

7

7

7

7

7

7

7

7

7

5

8

>

>

>

>

>

>

>

>

>

>

>

<

>

>

>

>

>

>

>

>

>

>

>

=

fqg 1

fqg 2

.

.

.

>

>

>

>

>

>

>

>

>

>

>

:

>

>

>

>

>

>

>

>

>

>

>

;

g

9

>

>

>

>

> fq g 1 >

c

8

>

>

>

>

>

>

>

>

>

>

>

>

f

q

n

>

>

>

>

<

>

>

>

>

>

>

>

>

>

>

>

:

fq g 2

.

.

.

fq g n

=

>

>

>

>

>

>

>

>

>

>

>

;

c

9

2

6

6

6

6

6

6

þ 6

6

6

6

6

4

c cc

c cq 1

c q 1 q 1

Symm

¼

8

>

>

>

>

>

>

>

>

>

>

>

>

>

<

>

>

>

>

>

>

>

>

>

>

>

>

>

:

F

c

F

F

.

.

.

q

q

F

q

1

2

n

9

>

>

>

>

>

>

>

>

>

>

>

>

>

=

>

>

>

>

>

>

>

>

>

>

>

>

>

;

:

c cq 2

0

c q 2 q 2

···

···

···

.

.

.

c cq n

0

0

0

c q n q n

3

7

7

7

7

7

7

7

7

7

7

7

5

8

>

>

>

>

>

>

>

>

>

>

>

<

>

>

>

>

>

>

>

>

>

>

>

:

_

c

fq_ g 1

fq_ g 2

.

.

.

f

q_

g

n

9

>

>

>

>

>

>

>

>

>

>

>

=

>

>

>

>

>

>

>

>

>

>

>

;

The entries of equations (12) are as follows:

ð 12Þ

n n

m cc ¼ J d þ X

J bi þ X

cos 2 b i f q g i ½I f q g i

i

m cq i ¼ cos b i

¼1

i

¼1

h ½ a i þR d b

2

1 þ c cos

2

i

b

i

m q i q i ¼ ½ I

i þ c cos 2 b i fq g i ½ I

n

c cc ¼ 2 X

i ¼ 1

T

cos 2 b i fq g ½ I f q_

i

g

i

c cq i ¼ 2 vc cos b i

c q i q i ¼ 2 c c cos 2 b i ½I

½ a i þR d ½ b i þ v cos 2 b i f q g i

_

n n

k cc ¼ k T 2 v 2 X J bi 2 v 2 X

i

¼1

i

¼1

cos 2 b i f q g i ½ I f q g i þ 2v cos 2 b i fq g i ½ I q_

f

g i

k q i q i ¼ EI i

r i L

4

i

k þ v 2

k

si

2 v 2 ð 1 þ c 2 Þ cos 2 b i ½I 2 2 v c cos 2 b i ð 1 þ cÞ½I

_

ð13 Þ

where fq g i is the i th blade modal vector that has a size equal to the number of modes N

is the torsional

is the i th blade

are one dimensional

vectors that represent the coupling between the i th blade modal degrees of freedom and the torsional degree of freedom of the system. The size of dynamic system, equations (12), depends on the number of blades n in the system and the number of modes N considered for each blade, which result in a system of ð1 þ n £ N Þ £ ð1 þ n £ N Þ size. The first matrix in equations (12) is the coupled inertia matrix, wherein its first entry is the torsional degree of freedom inertia that is

considered for this blade, b i is the setting angle for the blade,

excitation torque, F

stiffening matrix due to system rotation, the vectors ½a i , b

F c

q

i is the i th blade modal excitation vector,

i

k

si

Blades condition

monitoring

281

JQME

12,3

282

function of the disk inertia J d , the blades inertia J b , the setting angles b i and nonlinear

function of the blades elastic modal deflections fq g i . The rest of the first row entries of the inertia matrix are the coupling vectors between the torsional degree of freedom and the blades modal degrees of freedom that are functions of the setting angle b i , the coefficient

vectors ½a i and b i and the torsional deflection c multiplied by the blades modal degrees of freedom f q g i . The diagonal entries are the modal inertia matrices for each blade. The second matrix is a matrix that developed from the formulation, not damping matrix, and can be called the Coriolis matrix. The entries of this matrix are nonlinear functions of the torsional deflection and velocity, the blades modal deflections and velocities, the setting angles and the speed of rotation. Structural damping can be added to the entries of the second matrix at corresponding blades entries. The third matrix is the equivalent stiffness matrix with the first entry corresponding to the torsional degree of freedom stiffness. As shown, in addition to the shaft torsional stiffness k T , the

equivalent torsional stiffness is softened as a result of speed of rotation v, the blades inertia J b , the setting angles and the blades modal deflections. It is stiffened as function of the blades modal deflections and velocities that are multiplied by the speed of rotation. The diagonal entries of the stiffness matrix are the blades modal stiffness matrix that are softened and stiffened as function of the speed of rotation, the setting angles and the torsional deflection. Moreover, the inclusion of the stiffening effect in the form of the

has produced coupling in the stiffness matrix between the individual modes

matrix k

s

i

of each blade; as the matrix

k

s

i is non-diagonal.

3. Numerical simulations and discussion To simulate the dynamic model of equations (12), the data of the experimental set-up available in the Advanced Mechanics Laboratory (AML) at KFUPM are used. The properties of this three-blades-disk-shaft system, shown in Figure 2, are given in Table I. Structural damping is accounted for in the simulation by adding the Rayleigh proportional damping to the system with damping ratio j ¼ 0: 01 for the shaft torsional and blade bending modes as all are made of steel. The blades undamped natural frequencies, for the non-rotating and uncoupled system, are calculated as shown in Table II. In all simulations, the system basic parameters are kept the same except the parameter that is explicitly mentioned to be changing.

 

Property

Value

Blade

material

Steel ( E ¼ 200 GPA, r^ ¼ 7 ; 850 kg = m 3 ) 0.125 m 2 : 54 £ 0 : 1 cm 0.2 kg/m 0 : 4233 N: m 2 0.05 m 3.08 –kg 0 : 01233125 kg :m 2 G ¼ 80 GPA 0.5 m 1 cm 157.08 N.m/Rad

Blade

length L

cross Blade mass per unit length, r

Blade

section

Blade

flexural rigidity, EI

Disk radius, R D Disk mass M d Disk moment of inertia, J D

Steel

shaft

Shaft

length

Table I.

Shaft

diameter

Blade-disk-shaft data

Torsional stiffness k T

The dynamic model, equations (12) has a size of 16 £ 16 for the considered three-blades model with five modes for each blade, is solved using the ODE15S solver of the MALTAB package and the system free and forced vibration response is monitored. The system response is represented as direct time history and power spectral density (PSD). The PSD gives the amplitude at each frequency component squared and divided by the frequency bandwidth (Hz) to eliminate the spectrum amplitude dependence on the frequency bandwidth (Thompson and Dahleh, 1998). The simulations are conducted for free and forced vibration of the blades.

Blades condition monitoring

283

3.1. Free vibration simulations For free vibration, selected blades are given static tip deflections d and the individual corresponding modal values are calculated based on the integration of assumed static deflection curve that is physically acceptable. After integration, the modal initial conditions for the first five modes as function of the blade i tip deflection are found in the form:

q i ð 0Þ ¼ d i

ffiffiffiffiffiffiffiffi

p

r i L i

2

0: 9707 20 : 0247 0: 0046 20 : 0008 0 : 0003

:

ð14 Þ

For the non-rotating three-blades unit with properties given in Table I and setting angles b ¼ 0 8 for all blades, a free vibration simulation is conducted. For the purpose of introducing mistuning, blade 2 length is made 0.9 of other blades length and blades 1 and 2 are given tip deflections of 2 0.012 m. The system response is shown as time history for blades 1-3 and shaft torsional vibration in Figure 3. The difference in the period of oscillation between Figures 3a and 3b, for blades 1 and 2, shows the effect of blade 2 having shorter length. Blade 3 was not given any initial tip deflection, however its time history, Figure 3c, shows more sophisticated time history with less decaying rate. The torsional vibration signal, Figure 3d, shows also similar time history as blade 3. The corresponding frequency spectrums are shown in Figure 4. Figure 4a shows that blade 1 is vibrating at its own first mode natural frequency of about 52 Hz. Blade 2 frequency spectrum, Figure 4b, contains its first mode natural frequency, which is about 60 Hz due to the decrease in its length by 10 percent. Although blade 3 is not given any initial excitation, it vibrates with three different frequencies as shown in the spectrum of Figure 4c. The 52 Hz and 60 Hz frequencies are blades 1 and 2 vibration frequencies, respectively. The third frequency component is 325 Hz, which is blade 3 second mode free vibration frequency due to its base excitation by the shaft torsional vibration. This means that blades 1 and 2 free vibration excited the shaft torsional vibration and thus excited blade 3 vibration. Figure 4d shows the torsional vibration frequency spectrum associated with blades free vibration. The spectrum shows not

Degree of freedom

Natural frequency (Hz)

Blade 1st bending mode

52.1

Blade 2nd bending mode

326.25

Blade 3rd bending mode

914.271

Table II.

Blade 4th bending mode

1,791.6

Uncoupled blades natural

Blade 5th bending mode

2,961.6

frequencies

JQME

12,3

284

Figure 3. Shaft-disk-blade free vibration response time history, blades 1 and 2 are given initial tip deflection of 2 0.01m

3. Shaft-disk-blade free vibration response time history, blades 1 and 2 are given initial tip deflection
Blades condition monitoring 285 Figure 3.

Blades condition monitoring

285

Figure 3.

JQME

12,3

286

Figure 4.

Shaft-disk-blade free

vibration response

frequenct spectrums

JQME 12,3 286 Figure 4. Shaft-disk-blade free vibration response frequenct spectrums
Blades condition monitoring 287 Figure 4.

Blades condition monitoring

287

Figure 4.

JQME

12,3

288

Figure 5. Shaft torsional vibration frequency spectrum for a rotor rotating at speed v ¼ 100 Hz

only blades 1 and 2 free vibration frequency components but also captures blade-3 second mode free vibration frequency, which is excited by the shaft torsional vibration. In addition, one can observe that the difference in blades 1 and 2 vibration frequency is reflected in the torsional vibration spectrum. This directs the attention towards the broadness of torsional vibration spectrum as indication of mistuning in the blades.

3.2. Forced vibration simulations To simulate operational condition the three-blades disk shaft model is operated at 100 Hz rotating speed and the shaft torsional vibration signals are monitored for different excitations and system properties. Figure 5 shows the torsional vibration spectrum when all blades have setting angles b ¼ 0 8 and blades 1, 2 and 3 are excited by sinusoidal force at their tips with frequencies 1 £ (at running speed), 2 £ (twice running speed) and 3 £ (three times running speed), respectively. As shown the torsional vibration spectrum captures the vibration of the three blades at their respective external excitation frequencies. In addition, the spectrum captures the blades free vibration frequency of about 70 Hz that is increased in this case due to stiffening effect. For the 100 Hz speed of rotation, the system is simulated when the setting angels are b 1 ¼ 0 8 , b 2 ¼ 45 8 and b 3 ¼ 85 8 under 1 £ , 2 £ and 3 £ excitation frequency for blades 1, 2 and 3, respectively. The torsional vibration spectrum, Figure 6, reflects the effect of setting angle by showing less sensitivity to blade 3 vibrations as b 3 ¼ 85 8 . The effect of shaft torsional rigidity on the sensitivity of the system torsional vibration signal to blades vibration is investigated by increasing the shaft torsional stiffness by 10 3 and 10 6 times its original k T . The torsional vibration spectrums, Figures 7a and 7b, show that the torsional vibration signal carries the

k T . The torsional vibration spectrums, Figures 7a and 7b, show that the torsional vibration
Blades condition monitoring 289 Figure 6. Shaft torsional vibration frequency spectrum for a rotor rotating

Blades condition monitoring

289

Figure 6. Shaft torsional vibration frequency spectrum for a rotor rotating at speed v ¼ 100 Hz

three-blades vibration signatures but with less amplitude as the shaft torsional rigidity is increased to extreme values approaching rigid shaft. The effect of increasing the drive mass moment of inertia is investigated by increasing the disk mass m d by 100 and 1,000 times. The resulting torsional vibration spectrums are shown in Figures 8a and 8b, for 100m d and 1,000 m d disk mass, respectively. As shown the spectrums capture the blades forced vibration signatures effectively. To this end one can observe that for all forced vibration simulations of the rotating system, the torsional vibration signal captured the blades vibration signatures with varying sensitivity depending on blade(s) setting angle, the amplitude of excitation, the shaft torsional rigidity and the drive mass moment of inertia. The least sensitivity is shown when the setting angle approaches 90 8 and the shaft torsional rigidity is infinitely high (approaching rigid).

4. Conclusions To investigate the feasibility of extracting rotating blades vibration from the shaft torsional vibration measurement, a mathematical model is developed and simulated in this study. The model considers n-blades attached radially to a rigid disk with setting angles, the disk is driven by shaft that is flexible in torsion and the shaft-disk blades system is rotating at constant speed. The Lagrangian dynamics is employed in deriving the equations of motion in conjunction with the assumed modes method to descritize blades flexibility. The model accounted for the blades axial shortening due to blades bending deflection in the form of the associated stiffening effect. The model is nonlinear and coupled second order ordinary differential equations with a size

JQME

12,3

290

Figure 7. Shaft torsional vibration frequency spectrum for a rotor rotating at speed v ¼ 100 Hz

JQME 12,3 290 Figure 7. Shaft torsional vibration frequency spectrum for a rotor rotating at speed
Blades condition monitoring 291 Figure 8. Shaft torsional vibration frequency spectrum for a rotor rotating

Blades condition monitoring

291

Figure 8. Shaft torsional vibration frequency spectrum for a rotor rotating at speed v ¼ 100 Hz

JQME

12,3

292

depending on the number of blades and the number of bending modes for each blade considered. The blades free vibration simulation results showed that the shaft torsional vibration signal carries the blades vibration signatures at their respective natural frequencies. In addition the effect in blades natural frequencies due to difference in properties was captured by the torsional vibration spectrum. Less sensitivity is shown for blades with setting angle approaching 90 8. The forced vibration simulations of the rotating system showed that the torsional vibration spectrum contains the individual blades forced vibration frequencies with magnitude depending on the setting angle, the amplitude of blades’ vibration, the shaft torsional rigidity and the drive inertia. The proposed mathematical model and simulation results showed that the approach of measuring the rotating blades vibration using the shaft torsional vibration measurement is feasible provided that sensitive and reliable torsional vibration pickup is used.

References Al-Bedoor, B.O. (1999), “Dynamic model of coupled shaft torsional and blade bending in rotors”, Computer Methods in Applied Mechanics and Engineering , Vol. 169, pp. 177-90. Al-Bedoor, B.O. (2001), “Reduced-order nonlinear dynamic model of coupled shaft-torsional and blade-bending vibrations in rotors”, ASME Journal of Engineering for Gas Turbines and Power , Vol. 123 No. 1, pp. 82-8. Al-Bedoor, B.O. (2002), “Blade vibration measurement in turbo-machinery: the current status”, The Shock and Vibration Digest, Vol. 34 6, November, pp. 455-61. Al-Bedoor, B.O. and Al-Nassar, Y.N. (2001), “ANSYS finite element analysis of multi-bladed rotors with flexible shafts”, King Fahd University of Petroleum and Minerals, unpublished work. Cookson, R.A. and Bandyopadhyay, P. (1980), “A fibre-optic Laser-Doppler probe for vibration analysis of rotating machines”, Transactions of ASME, Journal of Power , Vol. 102, July, pp. 607-12. Fan, Y.C., Ju, M.S. and Tsuei, Y.G. (1994), “Experimental study on vibration of a rotating blade”, Transactions of the ASME, Journal of Engineering for Gas Turbines and Power , Vol. 116, July, pp. 672-7. Kaufman, P. and Kershisnik, M. (1984), “Case history – a high speed turbine wheel”, Mechanical Engineering , July, pp. 38-45. Laws, W.C. and Muszynska, A. (1987), “Periodic and continuous vibration monitoring for preventive/predictive maintenance of rotating machinery”, Transactions of the ASME, Journal of Engineering for Gas Turbine and Power , Vol. 109, April, pp. 159-67. Maynard, K.P. and Trethewey, M. (1999), “On the feasibility of blade crack detection through torsional vibration measurements”, Proceedings of the 53rd Meeting of the Society for Machinery Failure Prevention Technology, Virginia Beach, VA, April 19-22 , pp. 451-9. Maynard, K.P. and Trethewey, M. (2001), “Application of torsional vibration measurement to blade and shaft crack detection in operating machinery”, Proceedings of the Maintenance and Reliability Conference Gatlinburg, Tennessee, May 6-9 . Maynard, K.P., Lebold, M., Groover, C. and Trethewey, M. (2000), “Application of double re-sampling to shaft torsional vibration measurement for the detection of blade natural frequencies”, Proceedings of the 54th Meeting of the Society for Machinery Failure Prevention Technology, Virginia Beach, VA, May 1-4 , pp. 87-94.

Muszynska, A., Goldman, P. and Bently, D.E. (1992), “Torsional/lateral vibration cross-coupled responses due to shaft anisotropy: a new tool in shaft crack detection”, Proceedings of the IMechE Conference on Vibrations in Rotating Machinery, C 432-090, Bath, UK , pp. 257-62. Nava, P., Paone, N., Rossi, G.L. and Tomasinin, E.P. (1994), “Design and experimental characterisation of a non-intrusive measurement system of rotating blade vibration”, Transactions of the ASME, Journal of Engineering for Gas Turbines and Power , Vol. 116, July, pp. 657-62. Reihardt, A.K., Kadambi, J.R. and Quinn, R.D. (1995), “Laser vibrometery measurements of rotating blade vibrations”, Transactions of the ASME, Journal of Engineering for Gas Turbines and Power , Vol. 117, July, pp. 484-8. Scalzo, A.J., Allen, J.M. and Antos, R.J. (1986), “Analysis and solution of a non-synchronous vibration problem in the last row turbine blade of a large industrial turbine”, Transaction of the ASME, Journal of Engineering for Gas Turbines and Power , Vol. 108, October, pp. 591-8. Simmons, H.R. and Smalley, A.J. (1990), “Effective tools for diagnosing elusive turbo-machinery dynamics problems in the field”, Transactions of the ASME, Journal of Engineering for Gas Turbines and Power , Vol. 112, October, pp. 470-6. Srinivasan, A.V. and Cuts, D.G. (1984), “Measurement of the relative vibratory motion at the shroud interfaces of a fan”, ASME Journal of Vibration, Acoustics, Stress and Reliability in Design , Vol. 106, April, pp. 189-97. Thompson, W.T. and Dahleh, M.D. (1998), Theory of Vibration with Applications , 5th ed., Prentice-Hall, Englewood Cliffs, NJ. Vance, J.M. (1988), Rotor Dynamics of Turbo-machinery , John Wiley & Sons, Chichester.

Blades condition monitoring

293

Corresponding author B.O. Al-Bedoor can be contacted at: albedoor@ju.edu.jo or bobedoor@kfupm.edu.sa

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints