Академический Документы
Профессиональный Документы
Культура Документы
Abstract
This paper presents a detailed seismic analysis of a powerful high-speed Russian turbine within a Nuclear Power Plant. Although dozens of
these turbines have worked reliably since the 1970s worldwide, until the last decade, only simplified structural analyses were available due to
the turbines’ complicated overall structure and internal design. The current analysis considers the detailed geometry of the turbine itself and the
vibration and seismic isolation system within the turbine’s pedestal and the full range of operational, accident and seismic loads like high pressure,
outside loads induced by pipelines and so on.
To solve the problem of the turbine seismic qualification, the following steps have been taken. The first step was to create detailed finite
element models of the turbine’s high and low pressure parts and rotor system with bearings. Using such models, corresponding simplified models
were developed to be included into the coupled model of the system: “Building–Vibroisolation Pedestal–Turbine” (BVT). The second step was
the analysis of that coupled system. Soil–structure interaction was considered using actual soil conditions. Three components of time history
acceleration were used to define seismic excitation. As the result of BVT system analysis, a full picture of time history displacements and loads
was determined. At the same time, a problem of rotor gaps was solved. In the final step, determined loads were applied to the detailed models of
the turbine’s parts for seismic qualification of the whole structure.
© 2006 Elsevier B.V. All rights reserved.
0029-5493/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2006.10.004
1316 V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324
The detailed spatial models were developed for all main tur-
bine parts. At the same time, the analysis of dynamic properties
(inertia, eigenvalues and stiffness) of these models allowed us
to create corresponding simplified models to be included in the
building model and correct their characteristics.
The high-pressure chamber of a turbine is a massive steel
casting device with high pressure inside. It has an approximately
symmetrical structure of about two planes. The first one is ver-
tical and goes through the axis of the turbine rotor. The second
one is normal to the axis of the turbine rotor. The high-pressure
chamber has 16 nozzles. Four inlet nozzles supply the device
Fig. 3. Most simplified models of the turbine and vibroisolation system. with high-pressure dry steam. The other 12 nozzles are for sepa-
ration and outlet of steam. With the help of CAD-CAE technique,
a detailed model of this device was developed. It used finite ele-
into four parts: turbine section, power supply section, oil sup- ments of SOLID type in general. The simplified model was also
ply section and generator section. That is why the building created in accordance with the detailed one (see Figs. 5 and 6).
constructively consists of cast reinforced concrete parts as The considered turbine has three identical low-pressure
well as a concrete skeleton with metal frame roof. Because chambers supported on the foundation slabs. Each chamber
these parts are essentially different in their stiffness and iner- is weld construction made of sheet steel with massive flanges
tia properties, the building is not symmetrical and regular. and diaphragms with blades. It has a middle part with double
Further, some kinds of twisting or torsion of the build- walls and outlet parts. The pressure of the steam drops from
ing may be expected. Therefore, only the detailed model of
the building can describe deliberately its behavior under an
earthquake.
At this stage of the analysis the extremely simplified mod-
els of equipment were coupled to the model of the building.
It was done only for equipment with mass which is compara-
tive to the building’s mass (>1%) (ASCE, 1998): turbine with
vibroisolation system, deaerator, etc. (see Fig. 3).
Direct linear time history analysis was carried out with con-
sideration of three samples of soil to develop broaden envelope
response spectra for the building base mat (see Fig. 4). Three
components of time history acceleration were synthesized to
be compatible with these spectra under the requirements of US
NRC (1989). This allowed us to exclude consideration of the
variation of soil properties and lead to one variant of further
analysis.
Fig. 4. Seismic broaden envelope spectra for building base mat (1: synthesized-
Y; 2: synthesized-X; 3: synthesized-Z; 4: horizontal X&Y-ground spectrum; 5:
Z-ground spectrum) (damping 5%). Fig. 6. Simplified model of the turbine high-pressure chamber.
1318 V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324
Fig. 8. Simplified model of the turbine low-pressure chamber. Fig. 10. Simplified model of the condenser.
V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324 1319
Fig. 11. Model of the rotor (A: high pressure part; B–D: low pressure part; E: generator rotor; F: exciter rotor; G: thrust bearing).
els based on the spring supports and constrained on the edges of viscous-elastic members of GERB GmbH (Germany). The
connection nozzles. vibroisolation is placed between horizontal concrete beams of
The turbine rotor lies on the 11 journal-bearings placed in the buildings and the frame of the pedestal. The model of the
supports’ cases on the cross fundament beams outside of high- vibroisolation pedestal includes BEAM and SHELL elements
pressure and low-pressure chambers. Stiffness of the rotor was to describe the frame and SPRING elements with concentrated
modeled with the help of BEAM elements with suitable cross- damping for vibroisolation (see Fig. 13). Namely viscous-elastic
sections (see Fig. 11). At the same time, the elastic modulus members were modeled by two parallel Maxwell’s series of
varies from section to section depending on temperature. Mass elastic spring and ideal viscous damper (Kostarev et al., 1994).
and inertia of the rotor and coupled features were described using To consider dynamic properties, the eigenvalues and mode
concentrated masses and inertial elements (MASS elements). shapes were determined for the coupled “Vibroisolation
To avoid the longitudinal oscillations, the rotor is supported Pedestal–Turbine” (VPT) system with assumption of rigid sup-
by thrust bearing which is the most loaded part of this subsystem. porting structures. The linearized stiffness of thrust bearing was
Moreover, this bearing has essentially non-linear stiffness under used for it. It should be noted that bending mode shapes are
static load as well as under a dynamic one (see Fig. 12). excited in the case of non-uniform distribution mass, stiffness
The supports’ cases and coupled mass of underlying pedestal or damping in that system along the rotor axis (see Fig. 14).
beams were modeled with SHELL elements. These supports Therefore, it may lead to rising of seismic inertial loads to the
were connected to the rotor with the non-linear SPRING ele- support system of the low-pressure chambers. Moreover, the
ments which had stiffness of oil layer of the bearings. same effect can be caused by non-uniform stiffness of the under-
The vibroisolation pedestal includes a massive reinforced lying building beams which are the base of the vibroisolation
concrete frame and vibroisolation system with elastic and pedestal.
The main mode shapes of the VPT system correspond to the
oscillations along the rotor axis (X-axis), across it in the hori-
zontal plane (Y-axis) and in the vertical direction (Z-axis). The
eigenvalues of these shapes are 1.95 Hz (for X-axis oscillation),
2.05 Hz (for Y-axis oscillation) and 3.14 Hz (for Z-axis oscilla-
tion). Usually an engineer uses the simple oscillator model (mass
with stiffness) for the rough estimation. In our case the oscilla-
tion model of the VPT system has eigenvalues of 2.71 Hz (for
X-axis and Y-axis oscillations) and 3.70 Hz for (Z-axis oscil-
lation). These values are essentially different from the values
obtained above. Therefore, the dynamic properties of the VPT
system should be taken into account.
Fig. 14. “Vibroisolation Pedestal–Turbine” system. Mode shape no. 10 (3.18 Hz).
Fig. 16. Mode shape no. 1 of BVT system. Oscillation along the rotor axis (1.30 Hz).
V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324 1321
Fig. 17. Mode shape no. 2 of BVT system. Oscillation across the rotor axis (1.67 Hz).
Fig. 18. Mode shape no. 3 of BVT system. Oscillation in vertical direction (2.98 Hz).
Table 1
Differences in mode frequencies and seismic spectral accelerations for different models of VPT system
Type of the model Direction of oscillations/seismic load
X Y Z X Y Z
VPT incorporated in the building model 1.30 1.67 2.98 0.85 0.99 1.26
VPT on rigid base 1.95 2.05 3.14 1.19 1.23 1.43
Simple oscillator 2.71 2.71 3.70 1.58 1.63 1.99
1322 V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324
The eigenvalues of these modes are significantly lower than The time history of relative displacements of the rotor about
the values observed for that system above. They equal to 1.30 Hz the low-pressure chamber are shown below (see Figs. 19 and 20).
(for X-axis oscillation), 1.65 Hz (for Y-axis oscillation) and The analysis of the maximal absolute values of these displace-
2.98 Hz (for Z-axis oscillation). It means that the total stiffness ments leads to conclusions about capacity of the turbine under
of the vibroisolation system and pedestal is comparable to the the earthquake.
stiffness of the underlying building supports. The spectral accel-
erations for the considered models are shown in Table 1. They 6. Analysis of the turbine parts using detailed models
represent seismic loads acting on the turbine. The detailed model
values are sufficiently lower than other corresponding values. The loads obtained by seismic analysis of the whole BVT sys-
Therefore, the complex approach with the consideration of the tem were used in the analysis of each part of the turbine based on
BVT system is absolutely suitable and justified. It leads to the detailed three-dimensional models. The results of the particular
analysis of the high-pressure chamber, low-pressure chamber 3. Response spectra calculation to find deformation, stresses
and the fragment of the support under seismic and operational and reactions due to the inertial seismic loads.
loads are shown in Figs. 21–23.
Three following steps were done to perform such analysis: The results of these three steps were combined to evaluate
the summarized values.
1. Static analysis of the devices under operational loads (dead- It should be underlined that the application of the response
load and pipelines load). spectra method (RSM) for the huge models constitutes the sep-
2. Series of the static analyses under maximal seismic reactions arate problem. The method was developed for the simplified
of connected equipment and pipelines. models of structures with a few degrees of freedom in the mid-
dle of 20th century. Then, only models with a limited number of
springs and masses were available for the seismic analysis. First,
the in-structure response spectra were developed from the accel-
eration time history computed with the help of a stick model of
the building. Second, such spectra were applied to the very rough
and simplified models of the equipment inside the building. But
now, simplified models of buildings and equipment have been
replaced by extremely detailed spatial models with hundreds of
thousands or millions of DOF. So the implementation of RSM
needs overall verification. It has been done in a number of works.
First, the use of detailed spatial models of a building (like the
model of the turbine hall mentioned above) allows for more accu-
rate descriptions of its behavior. For example, coupled motion of
slabs and walls can be observed during an earthquake. An ana-
lyst can observe all six components of motion in the exact place
of the equipment installation with the help of such a detailed
model. But only maximal modal responses of the equipment on
each of the three translational components of motion are used
in RSM instead of their time history. The time moments when
these maximums happened are lost. Usually, two rules of com-
bining of three seismic responses on each global direction of
motion (X, Y and Z) are used in RSM. They are “square root
of sum of squares” (SRSS) and Newmark’s “100-40-40” rule.
Both of them are based on the assumption of uncorrelated com-
Fig. 23. Effective stresses for support element. ponents of in-structure motion. Actually, such an assumption is
1324 V.V. Kostarev et al. / Nuclear Engineering and Design 237 (2007) 1315–1324