Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The structural optimization technique is well known as one of the tools to make a better design. Recently,
Received 27 February 2009 it is applied to the design of nuclear components for the purpose of the design evolution and enhancement.
Received in revised form 3 April 2011 When it is applied to the nuclear components, there are a couple of points to be considered. The ASME
Accepted 27 April 2011
section III, design by analysis, provides stress criteria for nuclear components, which is quite different
from the general stress constraints for a typical optimization process. Based on the ASME section III,
stresses are categorized into three types with different stress limits. The stress limits change according
to corresponding service levels. Since the types of stresses and the service levels introduce different
stress, the multiple loading conditions should be utilized during an optimization. And the general stress
constraints during an optimization are the maximum nodal stresses applied to the concerned points
rather than the concerned section. On the contrary, the failure condition of the ASME section III can be
evaluated by comparing the stress intensity, which was created by the stress linearization procedure at
the concerned section, with the prescribed one. So, the linearized stresses such as stress intensities are
used as stress constraints during an optimization for nuclear components. To validate the ideas, the shape
of torus seal is optimized under the multiple loading conditions with the linearized stress constraints.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction (ASME, 2007). The stress limits change for different service levels.
Since these stress limits change for the types of stresses and the ser-
Since a structural optimization was introduced by Schmit vice levels, multiple loading conditions should be utilized during an
(Schmit, 1960) by using a finite element analysis, many researches optimization. According to the ASME code, the failure condition can
have been undertaken with the purpose of a design enhancement be evaluated by comparing the stress intensity, which was obtained
by applying on optimization technique to large-scale structures from the stress linearization at the concerned section, with the
especially in the industries. Since the object for an optimization is prescribed one. This prescription of the ASME code is somewhat
finite element model with many degrees of freedom, it is not easy to different from the general stress constraints during an optimiza-
find a mathematical optimum design. The objective of a structural tion. Stress constraints during an optimization are usually applied
optimization is to generate a better design in an engineering sense. to the points of interest rather than the concerned section. On the
According to the development of nuclear technology, new designs contrary, section-wise stresses should be used in the ASME code.
for nuclear-related structures and components are necessary. The So, linearized stresses are used as the stress constraints during an
optimization technique is considered as one of the tools to achieve optimization for nuclear-related structures. Choi et al. (2008) tried
a new better design. In this research, several points to be consid- to apply a linearized stress to a shape optimization procedure for
ered when the optimization technique is applied to nuclear-related a perforated pressure vessel. But this optimization was conducted
structures are discussed. And ways to manipulate these problems under one simple loading condition, which means that only one
are proposed. There are two considerations in optimizing nuclear- stress limit was applied. Thereafter, an optimization under sev-
related structures. One is a manipulation of the multiple loads with eral loading conditions with different stress limits has been studied
different stress limits. And the other is a peculiar procedure to cal- (Choi and Seo, 2008).
culate the stresses used as constraints. Based on the ASME code,
stresses are categorized into three types with different stress limits
2. Structural optimization to meet ASME code section III
0029-5493/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2011.04.046
2654 W.-S. Choi, K.-S. Seo / Nuclear Engineering and Design 241 (2011) 2653–2659
Table 2
Nomenclature Stress limits for each service limit (other than bolts in transport system).
Table 1
Stress limits for each service limit.
Primary membrane Pm ≤ Sm Pm ≤ 110% Sm for Level B Pm ≤ greater of 1.2Sm and Sy Pm ≤ 2.4Sm and 0.7Su
Primary bending PL + Pb ≤ 1.5Sm PL + Pb ≤ 110% (1.5Sm ) for PL + Pb ≤ greater of 1.8Sm and Elastic system analysis
Level B 1.5Sy —–OR—– PL + Pb ≤ (1.5)2.4Sm and
(1.5)0.7Su
PL + Pb ≤ 0.8CL-limit —–OR—–
Load ≤ 0.9CL-limit * or
0.9CL-plastic or test
*
yld = lesser 2.3Sm &
0.7Sy
Table 4
Stress limits for each service limit (storage system).
Stress category All service limits Hydrostatically tested Pneumatically tested containments
containments
Stress intensity factor k is 1.0 for the design and Level A. For Level C, k is 1.2. For Level D, the stress limits of ASME code section III, Division 1, Appendix F shall be applied.
The stresses based on the ASME code are not point-wise stresses
but section-wise stresses. If the stress classification line is selected,
the location where the stress is evaluated is fixed. So, in this case, Eq.
(3) is used for the stress constraints. By this manipulation, the total
number of stress constraints treated in each iteration reduce and a Fig. 2. . The linearized stress component.
2656 W.-S. Choi, K.-S. Seo / Nuclear Engineering and Design 241 (2011) 2653–2659
Table 6
Loading combination.
1 (1) 1.5Sm
2 (1) + (2) 1.5Sm
3 (1) + (2) + (3) 3.0Sm
(m − I)V = 0
(b − I)V = 0
(9)
[(m + b ) − I]V = 0
[(m − b ) − I]V = 0
Fig. 3. . The flowchart of the proposed optimization.
Table 5
Stress category classified by the loading and boundary conditions.
Fig. 4. The shape of torus seal, loading/boundary condition, and design variables.
Table 7
Current design and the optimum design.
Rad (mm) 20 80
T (mm) 5 17.4
H1 (mm) 30 34.1
H2 (mm) 18 19.4
Delta (mm) 0 69.7
a certain range of design variables is performed, a sequential opti- In the optimization by using the commercial software in which
mization procedure under a different range of design variables is an approximation method is utilized, it is assumed that the design
iterated with the result of the previous optimization as a new start- reaches to an optimum, when the final design of the current
ing value. Several optimization procedures are necessary to find an optimization procedure equals to the final design of the previ-
optimum design. A design is assumed to be an optimum design if ous optimization procedure. In other words, when we perform
the difference between the current design and the previous design optimization with the final design of the previous optimization pro-
Fig. 6. Stress result for the initial design and the optimal design.
2658 W.-S. Choi, K.-S. Seo / Nuclear Engineering and Design 241 (2011) 2653–2659
cedure as the initial design, there is no further design enhancement, rication tolerance of a segment gate is applied as a displacement
which means the value of i equals to zero. The same convergence boundary condition. The imposed magnitude of relative displace-
criterion was applied to the example in chapter 3. The flowchart of ment was determined by evaluating installation tolerance. Since
the proposed optimization can be summarized and schematized in the materials of two connected structures are different to each
Fig. 3. other, the gap between two structures could be increased due to
the different thermal coefficients. To simulate this gap, the relative
displacement in horizontal direction is imposed as a displacement
3. Example
boundary condition. The magnitude of relative displacement was
determined by a heat transfer analysis. These individual loadings
A torus seal is selected as an example for an application of the
are expressed as (1), (2), and (3).
proposed optimization technique for a nuclear-related structure.
The loading combinations are in Table 6 with the allowable
Torus seal is usually used in a nuclear-related structure to pre-
stress limits. The loading combinations consisting of more than one
vent a leakage. The cross section of a torus seal is the shape of an
loading are expressed as #1, #2, and #3. The loading combinations
omega character. The shape of a torus seal selected as an example
are divided into three ones: the internal pressure only, the inter-
is shown in Fig. 4. A torus seal was evaluated as one of designs to
nal pressure and the displacement in vertical direction, and the
maintain a seal function of a separated reactor cover. When there
internal pressure and displacements in both of vertical and hori-
is a structural discontinuity, a leakage between two different struc-
zontal directions. The allowable stress limit is based on the ASME
tures should be prevented by welding each part of a torus seal onto
code regulation. Loading combination #3 has an allowable stress
different structures, respectively. In case those two parts should be
limit of 3.0 Sm because a thermal loading is included. These loading
separated, the removal of a torus seal can make it. The main func-
combinations are adopted as multiple loading conditions.
tion of a torus seal is to maintain sealing. In addition, an internal
For the stress constraints for the parametric optimization of the
pressure, relative displacements, and thermal loadings could occur
torus seal, linearized stresses based on ASME section III and dif-
at the inside of torus seal or the structural discontinuity region.
ferent stress limits according to stress category are adopted. In
The design pressure is applied to the torus seal as an internal pres-
order to adopt a stress linearization procedure to constraints dur-
sure. So, a torus seal accommodates an internal pressure. A relative
ing an optimization, an extra data manipulation such as calculating
displacement in the horizontal or the vertical direction happens
a linearized stress and a derivative of the linearized stress needs
because of a tolerance from a production and installation of it. This
to be conducted. Stress classification lines are established where
relative displacement can also happen because of a difference in the
the maximum stress gradient could occur and stress linearization
thermal expansion magnitude between two different structures.
for these SCLs is conducted. The seven selected stress classification
A torus seal has to maintain its function and structural integrity
lines (SCLs) are shown in Fig. 5.
under multiple loads such as an internal pressure, relative displace-
ments, and thermal loadings. For the design of a torus seal, several
parameters are defined to characterize its shape and a parametric
3.2. Results
shape optimization is performed by means of commercial software
(ANSYS, 2005).
The initial design and the optimum design of a torus seal are
shown in Fig. 6 and Table 7. The scale of stress contour in Fig. 6
3.1. Problem definition is matched for easy comparison. The initial size was chosen to be
excessively small to prevent a physical interface with other struc-
A finite element model for torus seal optimization and the design tural component. A preliminary analysis showed that the initial
variables are represented in Fig. 4. The radius and thickness of a design did not satisfy the given constraints. So, the initial design
torus seal were selected as design variables. The various heights to point of this optimization example started at the infeasible region.
characterize the shape of torus seal were also selected as design Consequently, the radius of the optimized design was bigger than
variables. The parameter of “delta” was selected to move the cen- the initial design in order to maintain the structural integrity under
ter of torus seal to the left or the right side. This movement was the given loading and boundary conditions. Generally, when an
estimated to contribute absorbing the vertical relative displace- internal pressure is applied to a cylinder structure, the magnitude
ment due to installation tolerance. The five design variables are of a stress is proportional to the internal pressure and the radius of
selected for the parametric optimization. The cost function is the the cylinder and inversely proportional to the thickness. In the case
outer radius of the torus seal. The radius is needed to be minimized of the torus seal, since the multiple loads and boundary conditions
to prevent an interface with other components. are mixed, an increment of the thickness can reduce the flexibil-
The characteristics of each loading, stress categories, and cor- ity. So, an optimization causes the radius and thickness to increase
respondent stress limits are represented in Fig. 4 and Table 5. simultaneously. The design variable “delta” increases to absorb the
The internal pressure acted on the inside of torus seal are applied displacement in the vertical direction coming from the installation
as a force condition during finite element analysis. The relative tolerance. The stress intensities and the allowable stress limit for
displacement in vertical direction due to the installation and fab- the initial design and the optimum design are shown in Fig. 7. In the
W.-S. Choi, K.-S. Seo / Nuclear Engineering and Design 241 (2011) 2653–2659 2659
middle figure of Fig. 7, the stress intensities for the initial design the constraints for each loading condition. The proposed method
are in excess of the allowable stress limit, which means loading was applied successfully to an optimization of NSSS components in
combination #2 is the dominant case. The stress intensities for the this study under multiple loading conditions and linearized stress
optimum design satisfy the allowable stress limit. constraints.
4. Conclusions References