1.

Unbalance
E.A. Ogbonnaya, et al, [70], in this work, Gas Turbine rotor shaft dynamic modeling will
be based on the speed and the force response due to unbalance. During the design stage
of Gas Turbine rotor shaft, the dynamic model is used to ensure that any potential
harmful resources are outside the engine operating speed. It is shown that when a system
is subjected to force harmonic excitation, its vibration response takes place at the same
frequency as that of the excitation. To determine the response of the shaft under
vibration, readings were collected from bearings 1, 2, and 3 of GT 17 in Afam thermal
Station as shown in table 1, while the engine characteristics are shown in Appendix A.
Equation 40 was developed to determine the response of the system under vibration. This
mathematical equation is used to run a computer program with a code in C+
+programming language.

Keivan Anbarani [71], in these experiment four eccentric masses is used in conjunction
with four springs and one damper to simulate the vibration. Masses are aligned in
different orders to simulate in-phase and out-phase situation. Acceleration, frequency and
phase shift of the system is measure, which then are used to do calculations and
generating graphs to describe the simple harmonic motion. He concluded that the
resonant frequency is the point that the system has maximum oscillation and in out of
phase case they have two natural frequencies, one around 7 Hz and other around 11 Hz.
The damping ratio at 7 Hz is much smaller because r is equal to 1. The source of error in
there theoretical versus experimental values includes friction, air drag and the damper
since they assume zero damping for the theoretical calculations. Channel two didn't show
a clear harmonic acceleration response because there are other external forces that could
affect the harmonic motion such as, air friction, internal friction for the motor and the
mass of the spring.

Jayanta Kumar Dutt, et al, [72], the main objective of this work is to determine reduction in
the unbalance response of a rotor shaft system by using a suitable polymeric or
viscoelastic bearing support. For analysis, a simple rotor system has been taken with the
rotor placed in the middle of a massless shaft with linear elastic bearings at the ends,
having viscoelastic supports. They concluded that it is possible to reduce the rotor
unbalance response over a wide frequency range. Desired frequency dependence of
support parameters β 1 and η may be obtained using the procedure outlined and an actual
viscoelastic material may be chosen with characteristics closest to the desired ones.

Viliam Fedák, et al, [73], they presented a serious engineering problem because it is a
major cause of excessive vibrations, at higher speeds. Arising large centrifugal
unbalanced forces can lead to damage of bearings and finally to destruction of machines.
The main goal of them simulation study consists in off-line analysis of influence of
rotating body eccentricity to its vibrations, their following elimination by calculation of
position and value of additional masses removing vibrations. They concluded that a
simulation model for analyzing vibrations of both static and dynamic unbalanced rotating
bodies has been developed. All calculations and simulations were performed in the
MATLAB/Simulink environment and the simulation model was involved into the GUI.
The analysis of simulation show, vibrations of the rotating body were considerably
decreased by adding the balancing masses on the place determined by the program. In
case of increasing the speed, the rotations of the rotor reaches instability region and
increased vibrations of longitudinal and transverse occurs there. They are caused by the
fact the rotor speed approaches to the resonance frequency.

B.Kiran Kumar, et al, [74], they performed experimental studies on a rotor to predict the
unbalance in rotor. The vibration velocities were measured at five different speeds using
FFT (Fast Fourier Transform) at initial condition. Based on vibration readings spectrum
analysis and phase analysis was carried out to determine the cause of high vibrations. By
observing the spectrum unbalance was identified. Then Rotor was balanced and found
that vibrations were reduced. The experimental frequency spectra were obtained for both
balanced and unbalanced condition under different unbalanced forces at different speed
conditions. They concluded that; as the speed increases the amplitude at 1X is also
increases for the same unbalance weight. This increase in amplitude value is because of
the different unbalanced force. Since the system frequency is nearer to 1000rpm due to
the presence of resonance at this speed higher amplitudes were presented.

Anindya Sundar Das, et al, [75], an investigation of active vibration control of an

unbalanced rotor–shaft system on moving bases is attempted in this work with
electromagnetic control force provided by an actuator consisting of four electromagnetic
exciters, placed on the stator in a suitable plane around the rotor–shaft. Numerical
simulations performed on the flexible rotor–shaft modeled using beam finite elements
shows that the control action is successful in avoiding the parametric instability,
postponing the instability due to internal material damping and reducing the rotor
response relative to the rigid base significantly, with sufficiently low demand of control
current in comparison with the bias current in the actuator coils. The conclusion of them
work is; Rotor–bearing system subject to base motion may experience instability due to
parametric excitation originating from time-dependent stiffness effect due to the
movement of the base in addition to the instability due to rotor–shaft internal damping.
Applying the control force from the actuator, stiffness of the rotor can be altered in such a
way that the zones of parametric instability are shifted away from the zones of excitation.

J.K. Sinhaa, et al, [76], they presented a method that can reliably estimate both the rotor
unbalance and misalignment from a single machine run-down. The method is
demonstrated using experimental data from a machine with two bearings and a flexible
coupling to the motor. A sensitivity analysis has also been carried out for the proposed
approach with perturbation errors in the rotor and bearing models, to confirm the
robustness of the method. They concluded that a sensitivity analysis for rotor unbalance
and misalignment estimation was performed by introducing errors in the parameters of
the rotor and fluid bearing models, to check the robustness of the proposed method. It has
been observed that the maximum error in estimated unbalance amplitude is usually less
than 45% for simultaneous random errors of 5% in the rotor model and 10% in the fluid
bearing models. In practice, the authors consider that the errors in the rotor model should
be much less than 5% for many machines. In general it has been observed that the
estimation method is far less sensitive to errors in the fluid bearing modeling compared to
the rotor modeling error, and rotor systems having fluid bearings are also less sensitive to
modeling errors, compared to systems having rigid bearings.

DaeYi Jung, et al, [77], they presented an approximate harmonic analytical solution to
describe the limit-cycle behavior in a three dimensional rigid rotor/automatic balancing
devices system. Essentially, the solutions presented here capture both in-plane transverse
deflection and out-plane tilting motion of the system under the limit-cycle condition.
Here the whirl speed of the ABD balancer masses is determined via the solution of a non-
linear characteristic equation. Also, based upon the limit-cycle solutions, the limit cycle
stability is assessed via a perturbation and Floquet analysis exploring three main
parameters; ABD balancer mass, ABD damping, and axial location of ABD along the
shaft. It is found that for certain combinations of ABD parameters and rotor speeds, the
non-synchronous limit-cycle can be made unstable thus guaranteeing global asymptotic
stability of the synchronous balanced condition. Specifically, as the ABD balancer mass
decreases and the damping increases, the influence of limit-cycle condition tends to be
less dominant and it is also found that the appropriate selection of relative location of
ABD on the rotor can be an effective way to avoid or prevent this unwanted
nonsynchronous behavior.

Shiyu Zhou, et al, [78], Vibration suppression of rotating machinery is an important

engineering problem. In this paper, a review of the research work performed in real-time
active balancing and active vibration control for rotating machinery, as well as the
research work on dynamic modeling and analysis techniques of rotor systems is
presented. The basic methodology and a brief assessment of major difficulties and future
research needs are also provided. They concluded that the major problem faced by the
active vibration control scheme is the use of a limited number of actuators to control an
infinite number of vibration modes. This research can also play a scientific foundation for
the design of an efficient and reliable generic adaptive control system. It is clear that
active balancing can suppress the imbalance- induced vibration. It is also clear that the
active balancing can improve product quality and improve the fatigue life of the machine
and cutting tools and, hence, reduce the system cost.

A.S. Das, et al, [79], they proposed an active vibration control scheme for controlling
transverse vibration of a rotor shaft due to unbalance and presented a theoretical study.
The technique uses electromagnetic exciters mounted on the stator at a plane, in general
away from the conventional support locations, around the rotor shaft for applying suitable
force of actuation over an air gap to control transverse vibration. From the this study they
concluded that: The electromagnetic exciters provide non-contact control force of
actuation which significantly reduces the rotor vibration due to unbalance and enhances
stability limit speed of the rotor-shaft system leading to safer operations at higher speeds.
The support forces are also sufficiently reduced leading to longer bearing life.

Bashir Asdaque, et al, [80], in this paper the procedures for the determination of the
deflection; slope, shear force and bending moment at the extremities of the shaft are
employed. Critical speeds or whirling frequency conditions are computed using transfer
matrix method (TMM). For particular shaft lengths, rotating speeds and shaft-profiles, the
response of the system is determined for the establishment of the dynamic characteristics.
They concluded that Shaft profiling is a commonly adopted design method enabling
system deflection, mass and inertia minimization. Transfer matrix method is particularly
used to ease the computations. This approach provides clear dynamic benefits whilst
reducing raw material cost, however, it introduces dynamic analysis complications. In
particular the establishment of the vibrational characteristics of profiled elements present
challenging problems.

Shivang Shekhar, et al, [81], they investigated and found that the Proportional and high
frequency band limited Derivative control action (abbreviated as PDhfbl) is much more
efficient than the former in terms of unbalanced response reduction and increment of
Stability Limit Speeds (SLS) of long rotor-shaft systems. They also established a
philosophical link between the PDhfbl control action and a multi-element mechanical
suspension model, or viscoelastic suspension model, reported to reduce response and
increase stability limit speed of rotors in literature. They concluded from the study that:
The proposed Proportional and high frequency band limited Derivative or PDhfbl control
law is much better than the PD control law used for actuators in increasing the Stability
Limit Speed (SLS) of a rotor-shaft system and reducing the Unbalanced Response
Amplitude. This study inspires the investigation with other multi-element viscoelastic
suspension models in search of a more efficient actuator control law.

Shuiping Yan, et al, [82], a shaft train in power plants is comprised of many journal
bearings supporting turbines, generator and exciter. They proposed two methods; sum of
unbalance responses and sum of modal magnification factors to analyze the vibration
sensitivity of a rotor system with many frequencies in the vicinity of the service speed.
They concluded that two methods are proposed to evaluate the vibration sensitivity to
unbalance for turbine generator shaft trains. Both methods include the influence of all
frequencies in the range from 0 to e.g. 130 % service speed. The frequencies close to the
nominal speed certainly contribute the most to sensitivity values. High sensitivity may
not indicate a real high vibration, it means only a bigger probability of occurring high
vibrations. The real vibration level on site depends also on the unbalance magnitudes and
their distribution.

Andrea Vania, et al, [83], they discussed that when a foundation natural frequency is close
to the machine operating speed, the forces caused by even almost negligible malfunctions
can cause very high vibration levels. This paper shows a case history in which a natural
frequency of a generator case, and that of an auxiliary shaft, significantly affected the
dynamic behavior of a power unit. The existence of these critical natural frequencies has
been confirmed through specific experimental tests, the results of which are shown in the
paper. They showed that the case history of a generator, the dynamic behavior of which
was considerably affected by some natural frequencies of the supporting structure, close
to the machine operating speed. Impact tests carried out on the generator case and
pedestals, as well as on the auxiliary shaft that connected this machine to a gas turbine,
have allowed some undesired natural frequencies to be identified in a critical frequency
range.
 NUMERICAL INVESTIGATION

The software used in the numerical work is ANSYS CFD, STAR-CCM+ and the system
designed in SOLIDWORKS.

Brief on STAR-CCM+
STAR-CCM+ can create and modify computer-aided design (CAD) models of the
geometry using a fully parametric 3D feature-based modeler, establish a bidirectional link
between STAR-CCM+ and CAD/product lifecycle management (PLM) software so that
geometry updates are automatically represented in the simulation, repair and defeature
imported CAD geometry using industry-leading surface wrapping and repair tools,
automatically creating a closed surface in as much detail as necessary to ensure the
fidelity of the results, automatically build a computational mesh on the geometry, no
matter how complex, using polyhedral or trimmed hexahedral cells, robustly create layers
of prismatic cells that allows to accurately capture boundary layers , substitute
components of the design as it evolves and deploy best practices with tools such as the
Simulation Assistant that guarantee consistency between simulations, and eliminate
analyst-to-analyst variation.

STAR-CCM+ is part of the Simcenter™ portfolio, a comprehensive suite of simulation

software and test solutions that help companies address the engineering challenges of
today’s complex products. Simcenter combines simulation and physical testing with
intelligent reporting and data analytics to produce digital twins that more accurately
predict product performance and drive innovation throughout all stages of the product
development process.

The model used in STAR-CCM+ is simplified as shown in figure (4.5, 4.6):

Steady state simulation is chosen because the problem is relatively simple and due to the
limited computational resources, unsteady solution will obtain fluctuating values of
moment with time plus, this approach is very time consuming and needs a powerful
workstation ( only a laptop is used for this analysis), the solver uses three-dimension
incompressible Reynolds-Averaged Navier-Stokes equation for a single-phase analysis
which simply obtains the time averaged state of the flow around the impeller, that is how
the “time-averaged” moment due to the flow around acting on the impeller is obtained,
the software calculates the moment around the impeller via the resolution of pressure and
shear forces exerted on the impeller by the flow.

The boundary conditions of the system are:

 Initially, the velocity is set to zero in all directions

 Gravity is set in the vertical direction downwards
  Velocity inlet boundary condition is assigned at the valve gate and velocity value
is obtained from the volumetric flow rate value
 The computational domain is divided into two regions as shown in figure
(4.19,4.20) :
a) Fixed domain : inlet and outlet runners
b) Rotating domain : for the impeller and the surrounding region , to simulate the
rotational motion of the impeller MRF “moving reference frame” is
implemented, this technique is used for steady state cases to simulate
rotational effects by incorporating related computational procedure, MRF is
an approximation of the “transient rotational motion with small transient
fluctuations”
 Pressure outlet boundary condition is assigned at the flow exit gate with pressure
value set to ambient.

Meshing
A mesh (Fig.4.21) is a discretized representation of a geometric domain. This domain can
include real-world geometry, its content, and its surrounding environment. The physical
space that solved within is called the simulation domain. Generating a mesh typically
involves creating a suitable simulation domain. There are two types of flow that require
different approaches when creating the simulation domain: internal flow, such as a flow
in a pipe, and external flow, such as the flow around and through a car.
The mesher in Simcenter STAR-CCM+ improves the mesh of a geometric object and
makes it ready for analysis. Some meshers generate volume cells, while other meshers
attempt to simplify and repair geometry.

Initial surface:
1. Flow domain mesh: 2. Rotating domain mesh
Number of faces: 2462 Number of faces: 2072
Number of edges: 827 Number of edges: 1400

Wrapped surface:
The surface wrapper used to provide a closed, manifold, non-intersecting surface when
starting from poor quality CAD data.
1. Flow domain mesh 2. Rotating domain mesh
Number of faces: 536060 Number of faces: 2072
Number of edges: 7282 Number of edges: 1400

Remeshed surface:
In order to improve the overall quality of an existing surface and optimize it for the
volume mesh models, the surface remesher used to re-triangulate the surface.
The remeshing is primarily based on a target edge length that that supplied and can also
include feature refinement that is based on curvature and surface proximity. Localized
refinement that is based on part surfaces or boundaries can also be included. Specific
surfaces or boundaries can also be omitted from the process so that the original
triangulation from the imported mesh can be preserved.
The surface remesher is typically used for remeshing surfaces that are output from the
surface wrapper and STL type data. As well as improving the surface for the volume
meshers it also aids the subsurface generator when the prism mesher option is selected.

1. Flow domain mesh: 2. Rotating domain mesh

Number of faces: 104226 Number of faces: 1057894
Number of edges: 2409 Number of edges: 27743

Volume mesh:
Simcenter STAR-CCM+ contains different types of meshing models that can be used to
generate a volume mesh starting from a suitably prepared surface mesh. The typed used
in this study is Trimmed “trimmed hexahedral cell shape based core mesh”.
The trimmed cell mesher provides a robust and efficient method of producing a high-
quality grid for both simple and complex mesh generation problems.
The prism layer mesh model is used with a core volume mesh to generate orthogonal
prismatic cells next to wall surfaces or boundaries. This layer of cells is necessary to
improve the accuracy of the flow solution.
1. Flow domain mesh 2. Rotating domain mesh
Number of cells: 152651 Number of cells: 1325907
Number of interior faces: 447192 Number of interior faces:
Number of vertices: 180668 3671464
Number of edges: zero Number of vertices: 1499686
Number of edges: zero

Simulating physics

Simcenter STAR-CCM+ simulations are built on numerical algorithms that solve relevant
laws of physics according to the conditions that the simulation defines.

Simcenter STAR-CCM+ contains a wide range of physics models and methods for the
simulation of single- and multi-phase fluid flow, heat transfer, turbulence, solid stress,
dynamic fluid body interaction, aeroacoustics, and related phenomena. These physics
models are all selected using a physics continuum.
A physics model in Simcenter STAR-CCM+ defines how a physical phenomenon in a
continuum is represented.

Essentially, physics models define the primary variables of the simulation (such as
pressure, temperature, and velocity) and what mathematical formulation is used to
generate the solution. An appropriate combination of models is necessary for the
complete definition of a physics continuum.

Models in Simcenter STAR-CCM+ have varying levels of complexity and functionality,

but their major purpose is to work with solvers to obtain a solution and to help present the
information to the users. The tasks of a typical model include making relevant field
functions available and placing initial conditions and reference values in its continuum.

In situations where transport equations are required, the model provides these equations
to the solvers. Therefore model properties often include coefficients relevant to the
transport equations, as well as parameters relevant to the discretization approach.

In any continuum, models are selected and changed using the Physics Model Selection
dialog, which assists in recommending models and in selecting the appropriate
combination of models.

The simulation of continua requires a bridge between what is understood in the material
world and the mathematical constructs of transport equation models. In the physical
world, we understand concepts such as “walls” or “porous media”. In the modeling
world, we think in terms of “boundary conditions” or “volumetric source terms”.

These concepts are bridged in STAR-CCM+ by formally defining types, conditions, and
values in such a way that there is no ambiguity. These three entities are used to provide to
the continuum models complete information for dealing with boundaries, regions, and
interfaces.

For steady-state simulations, the converged solution should be independent of the initial
field. However, the path to convergence, and hence the computational effort that is
required to reach convergence, is affected. Therefore, choose the initial conditions and
values judiciously, particularly when the physics is complex.

Each model requires sufficient information for the primary solution data for the primary
variables that are associated with the model to be set. In most cases, the initial conditions
are set up directly. For some models, such as turbulence, there is the option of specifying
the information in a more convenient form (turbulence intensity and turbulent viscosity
ratio instead of turbulent kinetic energy and turbulent dissipation rate).

Most fluid flows of engineering interest are characterized by irregularly fluctuating flow
quantities.
Often these fluctuations are at such small scales and high frequencies that resolving them
in time and space come at excessive computational costs. Instead of solving for the exact
governing equations of turbulent flows (Direct Numerical Simulation), it is less
expensive to solve for averaged or filtered quantities and approximate the impact of the
small fluctuating structures. Turbulence models provide different approaches for
modeling these structures.

The turbulence models that are implemented in Simcenter STAR-CCM+ can be

subdivided into two categories:

 Models that provide closure of the Reynolds-Averaged Navier-Stokes (RANS)

equations.
 Scale-resolving simulations that solve for large scales of turbulence and model
small-scale motions. These are Large Eddy Simulation (LES) and Detached Eddy
Simulation (DES).

Most simulations will rely on the first approach. The second approach (LES and DES) is
best used after carefully reviewing the applicable literature to gain confidence that the
grid resolution requirements can be met and that the computational costs incurred by
resolving the small time and length scales are indeed justified.

It is generally recognized that all existing turbulence models are approximate

representations of the physical phenomena of turbulence. The degree of approximation in
a given model depends on the nature of the flow to which it is applied. The
characterization of the circumstances which give rise to 'good' and 'bad' performance
must be based mainly on experience.

Reynolds-Averaged Navier-Stokes (RANS) Turbulence Models

 RANS equations are solved for continuity and momentum using K-Ɛ (K-epsilon)
model which is valid for turbulent core flows ( in correspondence to our case )
coupled with wall functions approach to resolve the boundary layer without the
need for high mesh resolution near the walls ( because of the computational
limitation )