You are on page 1of 14

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

net/publication/331443556

Topology optimization design for offshore platform jacket structure

Article in Applied Ocean Research · March 2019


DOI: 10.1016/j.apor.2019.01.003

CITATIONS READS

43 4,168

6 authors, including:

Xiaojie Tian Guijie Liu


Ocean University of China Ocean University of China
59 PUBLICATIONS 1,184 CITATIONS 152 PUBLICATIONS 1,362 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Guijie Liu on 22 June 2021.

The user has requested enhancement of the downloaded file.


Applied Ocean Research 84 (2019) 38–50

Contents lists available at ScienceDirect

Applied Ocean Research


journal homepage: www.elsevier.com/locate/apor

T
Topology optimization design for offshore platform jacket structure
a,b a a,b,⁎ a a,b c,⁎⁎
Xiaojie Tian , Qingyang Wang , Guijie Liu , Yunxiang Liu , Yingchun Xie , Wei Deng
a
Department of mechanical and electrical engineering, Ocean University of China, Qingdao, 266100, China
b
Key Laboratory of Ocean Engineering of Shangdong Province, Ocean University of China, Qingdao, 266100, China
c
Institute of Oceanographic Instrumentation, Qilu University of Technology (Shandong Academy of Sciences), Qingdao, 266061, China

A R TICL E INFO A BSTR A CT

Keywords: Offshore jacket platform is widely used as production or oil recovering platform in the shallow sea, and is also
Offshore platform applied to the offshore wind turbine supporting structure in the recent years. The jacket structures are normally
Jacket structure designed to be conservative and bulky according to various design codes. In this work, a structural optimization
Topology optimization design method for jacket platform structure has been developed based on topology optimization theory. The
Finite element analysis
topology optimization method is applicable at an early design stage, which can determine the initial structure
and force transmission path. The whole design space is chosen as design variables, and the goal is to maximize
the structural stiffness. A set of constraints based on multi-criteria design assessment is applied according to
standard requirements, which includes stress, deformation, vibration and design variable constraints. The op-
timization results are compared with the original platform for static performance, dynamic performance and
Ultimate Carrying Capacity (UCC). Results show that the optimized structure show a 13.7% reduction in the
global mass, 46.31% reduction in the maximum equivalent stress, and large ultimate carrying capacity ability
under the same environmental loads. It is demonstrated that the proposed topology optimization method is
capable of effectively determining the optimal design of jacket platform structures.

1. Introduction requirements. The new design structure should improve the structural
performance and reduce manufacturing cost. For the design and man-
As the foundation platform of ocean engineering, jacket platform is ufacture of traditional offshore platforms, platforms are usually de-
the main supporting part of offshore structures, which is generally signed to be conservative and bulky, in order to ensure the normal
working within the depth of 10 m to 200 m. Jacket platform is widely exploitation of resources and the safety of workers, which caused high
used as production or oil recovering platform in the shallow sea, and is costs during the process of manufacture and installation.
also applied to the offshore wind turbine supporting structure in the Lots of scholars researched optimization methods for offshore sup-
recent years [1–3]. It subjects to environmental loads during their porting structures. Feng et al. [4] proposed the shape optimization
lifetime; these loads are imposed on the platform through natural design for jacket platforms by seeking the nodal position and cross-
phenomena such as wind, current, wave, earthquake, snow and earth sectional dimensions to obtain the minimum weight of platform. Liu
movement. The design, analysis, and construction of offshore structures et al. [5] presented acceleration-oriented design optimization of ice-
compatible with the extreme offshore environmental conditions are the resistant jacket platforms. This method focused on the dynamic per-
most challenging and creative task. Compare to the land-based, offshore formance of the jacket platforms to achieve an economical and rational
structures have added complication of being placed in an ocean en- design. Yang et al. [6] used the reliability based design optimization for
vironment where hydrodynamic interaction effects and dynamic re- the tripod sub-structure of offshore wind turbines. A much higher re-
sponse become major considerations in their design. Therefore, a good liability model is obtained based on the Monte Carlo simulation. Lee
jacket platform structure is very important for its service life. et al. [7] also compared methods of reliability-based design optimiza-
As one of the earliest emerging offshore platforms, the structural tion and deterministic optimization for a monopole transition piece in
design of jacket platform has become more and more mature. However, an offshore wind turbine system. Chen and Fu [8] proposed a fuzzy
with the consideration of green economy and new marine energy de- approach to the lectotype optimization of offshore platforms. Chew
veloping in recent years, design of jacket platform has put forward new et al. [9] presented an analytical gradient-based method to optimize the


Corresponding author at: Department of mechanical and electrical engineering, Ocean University of China, Qingdao, 266100, China.
⁎⁎
Corresponding author at: Institute of Oceanographic Instrumentation, Qilu University of Technology (Shandong Academy of Sciences), Qingdao, 266061, China.
E-mail addresses: liuguijie039@163.com (G. Liu), dengwei_sdioi@163.com (W. Deng).

https://doi.org/10.1016/j.apor.2019.01.003

Available online 09 January 2019


Received 19 March 2018; Received in revised form 23 November 2018; Accepted 2 January 2019

0141-1187/ © 2019 Published by Elsevier Ltd.


X. Tian et al. Applied Ocean Research 84 (2019) 38–50

diameter and thickness of offshore wind jacket substructure. Gentils for jacket structure is introduced in Section 5. Section 6 evaluate and
et al. [10] developed an optimization method coupled parametric finite verify the structural performance of the topology design. Finally the
element analysis and genetic algorithm offshore support structure. This conclusions are given.
method can optimize the outer diameter and thickness of various
components simultaneously. These optimization methods are usually 2. Methodology
based on the original structure configuration, and only the size and
position of piles are adjusted a little. So the effect of optimization is not The solving method for topology optimization mainly contains
particularly significant. analytical method and numerical method. The analytical method began
For structural optimization, there are three major categories: size, with Maxwell, where he first performed the basic topology analysis of a
shape and topology optimization, and the process of optimization is stress-constrained minimum weight truss in 1854. In 1904, Michell [24]
gradually increasing from easy to difficult [11–13]. The optimum proposed the conditions for the optimal truss, which is known as Mi-
structure derived from sizing and shape optimization is only a result of chell criterion. It is also considered as a milestone for the theory of
the initial topology design. In other words, the structure from sizing and structural optimization design. Although Michell theory has been
shape optimization may not be the true optimum structure if the initial greatly developed, it is difficult to solve the problem by using the
topology design is not an optimal structure [14]. Finding a good analytical method. Therefore, the application of structural optimization
structural configuration before shape and sizing optimization is an obtained by the analytical method is extremely limited. Now numerical
important but difficult task and stresses the need for topology optimi- methods are usually used to solve the optimization problems.
zation. For most designs, the goal is mainly to maximize the structural
With the development of topology optimization technology in re- stiffness, so usually, the objective function of topology optimization is
cent years, especially the development of continuum structure optimi- the value of a linear function for the equilibrium displacement field.
zation, the speed and scale of optimization are greatly improved. This linear function is called compliance, and it is the inverse of a global
Topology optimization can be incorporated in a very early phase of the stiffness, so when compliance is minimized, the structure will attain the
product development process to propose complete a new design maximum stiffness. The objective of our topology optimization problem
[15,16]. It is a considerably practical aid in structural design as it is to determine the stiffest possible structure, or find the solution with
produces design solutions of high performance by finding the optimal minimum compliance, for a given space; by providing the volume
structural layout. Several methods have been developed for topology constraints. Therefore, the optimization problem in this study can be
optimization of continuum structures, such as the solid isotropic mi- stated as:
crostructure with penalization (SIMP) method, the evolutionary struc-
N
tural optimization (ESO) method, level set method and so on [17]. SIMP min: c (x ) = UT KU = e=1
(x e ) P ueT k 0 ue
x
method assumes isotropic materials and takes the element relative V (x )
density as the continuous design variable [18]. ESO method realize the subjectto: V0
=f
determination of void elements’ sensitivities rather than heuristic esti- KU = F
mation. This formulation claims the justification for optima output from 0 x min xi 1 (1)
a convergent and mesh-independent procedure [19]. Level set method
Where, U is the global displacement matrix, F is the global force matrix,
is the implicit representation of structural boundary by using the zero
K is the global stiffness matrix, ue and k0 is the displacement vector and
level set of higher order functional in the design domain [20]. Both the
stiffness matrix of element, x is the vector of design variable, xmin is the
SIMP and level set methods employ sensitivity analysis and a mathe-
minimum vector of relative density (its value is usually greater than
matical programming method such as MMA, and they have been suc-
zero to avoid singularities), N is total number of finite element dis-
cessfully used to solve the minimum mean compliance problem and the
cretized by the design area, p is the penalty factor, V(x) and V0 are the
compliant mechanism problems. It has been reported that SIMP, level
material volume and design volume respectively, and f is the specified
set method are computationally effective, but their convergence to the
volume ratio.
global optimum solution is not guaranteed [21].
Especially for the truss structure, it can be directly optimized
through the establishment of continuum model to obtain a reasonable 2.1. Topology optimization method
force path. Aiming at offshore engineering structures, Liu et al. estab-
lished the calculation model of Marine platform in ANSYS software, and SIMP method is based on the continuous variable between 0–1 and
carried out static, dynamic and optimized design of the platform under the elastic modulus of material. The topology optimization is accom-
different working conditions. The topological optimization with volume plished by removing the value of small variables. SIMP method can
as the objective function reduces the result by 8% [22]. Lu et al. con- associate the density value of design variable with the element elastic
sidered the dynamic performance of the jack-up platform and con- modulus by interpolation function. Material interpolation schemes
ducted multi-objective optimization based on the dynamic performance consider the design variables as a continuous variable between com-
of the pile leg. The results showed that the method had good con- plete void or fill, where, the value xmin in Eq. (1) is a practical lower
vergence [23]. This article uses this idea to design the jacket structure, bound of the element density introduced to avoid singularity in the
and to find an initial structural configuration for jacket structure numerical implementation.
foundation by maximizing its structural stiffness. It also can provide The mathematical models for SIMP is expressed in Eq. (2):
enough resistance for extreme design loads and increase its dynamic E p (xe ) = E min + xep (E 0 E min ) (2)
performance.
This work attempts to explore the applicability of topology opti- Where, Ep indicate the elastic modulus after interpolation. E0 and Emin
mization techniques in the field of jacket platform structural design; are the elastic modulus of solid material part and void part respectively.
demonstrating how these techniques can help to develop better solu-
tions and procedures while reducing, at the same time, the costs of 2.2. Solving method
design, deployment and manufacturing.
The paper is organized as follows. Section 2 describes the optimum To solve topological optimization problems, there are several dif-
methodology used in this paper. Section 3 introduces the characteristics ferent approaches such as Optimality Criteria (OC) method, Sequential
of the reference jacket platform model. Section 4 analyzes the en- linear programming (SLP) method, Method of Moving Asymptotes
vironmental load of jacket platform. The topology optimization design (MMA) method, and some others. OC method [25] is mainly based on

39
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

an inspired display of the variable update scheme renew design vari-


ables. It is typically used in situations where the number of global
constraints is much less than the number of design variables. MMA
method [26] transforms the implicit optimum problem into a series of
explicit simple convex approximation sub-problems by introducing
moving asymptotic parameters. It has better adaptability for complex
topological optimization problems and is more suitable for the optimum
problem with multiple constraints and complex objective functions.
Considering the efficiency and robustness of algorithm, MMA
method is used in this paper to solve the continuum topology optimi-
zation problem. The optimization problem can be approximated by
MMA as shown in Eq. (3).

minf˜0 (x (k) )
x
(MMA) s. t . f˜i (x (k) ) 0
(k ) (k )
xj (j = 1, 2, ..., n; i = 1, 2, ..., m)
j j (3)

Where, α(k) j and β(k) j are the moving limits, fi is the approx-
imating function, the specific meaning of each parameter in the formula
is detailed in [27].

2.3. Numerical instability and suppression techniques of topological


optimization
Fig. 1. Offshore jacket platform photo in site.
During the process of topology optimization, there are some nu-
merical instabilities, such as the checkerboard pattern, mesh-dependent stability of the structure. The leg column, pile, and braces act together
optimization defects, which cause some difficulties for the topological to form the support structure for the jacket platform [29].
configuration extraction [28]. In order to get the desired structure to- In this study, JZ20-2MUQ platform is used as a reference. JZ20-
pology with clear, homogeneous and easily manufacturing properties, it 2MUQ platform is located at Liaodong Bay, Bohai Sea. It is a typical
is necessary to restrain the numerical instability. The main method is to jacket platform with four-legs. The designed water depth is 25 m, the
use a more stable finite element model, and filter the density or element overall height of the jacket is 40.5 m, the pile diameter is 1 m, the wall
sensitivity, etc. thickness is 0.02 m, and the weight of the upper deck is 1600 t. The
Sensitivity filtering method is used in this study, which is based on bottom size of the platform is 11 × 11 m2, and the top area is
convolution filtering. It is a local constraint method and centered on an 6.75 × 6.75 m2 (Fig. 1).
element. The sensitivity of this element can be replaced by the weighted Based on the above geometric dimensions, the 3D geometry of the
average value of each element sensitivities within its filter radius. The referenced model is generated, mainly consisting of the lower sup-
independence of mesh is ensured by changing the sensitivity. porting structure. The upper part is simplified as mass points acted on
N
the lower part. The sizes of main support structures are shown in
c 1 c Table 1. And the whole finite element model is established in Fig. 2.
= N
Hˆ f x f
xe xe Hˆ
f =1 f f =1
xf
(4) Usually, jacket structure is mainly composed of X-type and K-type
structure. The jacket in this study is welded by a four-layer K-tube
Where, Ĥf is the convolution operator (weight factor) and can be ex- structure and four X-tube structures.
pressed as: Because leg column, pile and braces are tubular structures, they can
Hˆ f = rmin dist(e, f ), {f N |dist(e, f ) rmin }, e = 1, ...N be effectively and accurately modelled using pipe elements. The ele-
(5)
ment type used here is PIPE59, which has six degrees of freedom at each
The control function dist(e,f) is defined as the distance between the node. And it is a uniaxial element with tension-compression, torsion,
center e of the element and the center f of the element. Outside of the and bending capabilities. Additionally, the element loads include the
filter area, the convolution operator Ĥf is 0. The convolution operator hydrodynamic and buoyant effects of the water and the element mass
decreases linearly with the distance of element f. include the added mass of the water and the pipe internals. Therefore,
PIPE59 element is very appreciable for the simulating pipe members
3. Reference jacket platform model and pile of the jacket below sea level.
The main part of the support structure is made of steel Q235, which
Jacket platform is mainly composed of two parts. One part mainly has been widely used for jacket platform structures [29]. Q235 is
contains the deck, and other facilities and equipment, which is called adopted for finite element model and its material properties are listed in
the upper structures. It can be used for oil and gas collecting and pro- Table 2.
cessing site, human life site and other uses. The other part is composed The coordinate system is set at the water surface and the bottom in
of jacket and steel pipe piles, commonly known as the support struc- four corners are mud surfaces. The upper section is represented as a
ture. It can be used to support the upper facilities and equipment. The lumped mass, applied via multi-point constraint on the model top. The
lower piles can support all loads and fix platform position, and its bottom is applied fixed constraints, while the effect of platform legs
number, length, and diameter are determined by the geological con- with soil is equivalent as a fixed constraint. Loads applied on the jacket
ditions and load of seabed. Jacket is a space frame composed of leg structure for both ultimate and operational conditions are defined in
columns and braces. Leg columns are hollow, and the steel pipe piles Section 4.
penetrate into the sea bed through the legs to fix the jacket and transmit
the upper load. The vertical and horizontal braces between leg columns
are called supports for transmitting loads and making sure the overall

40
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Table 1
Offshore platform parameters and element type.
Category Structure Outer diameter/m Thickness/m Element

1 Tilt brace for jacket 0.6 0.02 PIPE59


2 Horizontal brace for jacket 0.6 0.02 PIPE59
3 Main leg column 1.0 0.02 PIPE59

P = ( /2) u2Cs (7)


Where, F is the wind force, P is the wind pressure, ρ is the mass density
of air (1.22 kg/m3 for standard temperature and pressure), u is the wind
speed, CS is the shape coefficient, A is the area of object.
In the absence of data indicating otherwise, the shape coefficients in
Table4 are recommended for perpendicular wind approach angles with
respect to each projected area. In this paper, considering the shape of
the initial optimization model, we take the shape coefficient in 1.0.

4.2. Current load

In general, the speed of ocean current varies very slowly with time.
In engineering design, the ocean current is often regarded as a steady
flow for convenient calculation. And the force acting on the object is
merely dragged force. However, in the co-existence of waves and cur-
rents conditions, we cannot ignore the speed caused by the waves. At
this time, the calculation of current load should take into account the
drag force caused by superposition of the velocity of the current and the
wave water particle. The drag force load per unit area in the water is
shown in Eq. (8)
1
Pl = CD Uc2 (Pa)
2 (8)
Where, CD is the drag force coefficient, the size often takes 1 [32], ρ is
the seawater density and its value is 1025 kg/m3, UC is current velocity.
Fig. 2. Model of JZ20-2MUQ platform structure.
4.3. Wave load
Table 2
Material properties of Q235. For the small-scale object, it is defined as D/L ≤ 0.2, where D is the
Property Symbol Value
characteristic length of an object and L is the wavelength. For example,
D is the diameter for a cylinder. For shallow waves, the wavelength L is
Young's modulus E 2.1 × 1011 N·m−2 [33]:
Poisson ratio μ 0.3
Shear modulus τ 7.6 × 1011 N·m−2 L = T gh (9)
Material density ρ 7800 Kg·m−3
Allowable normal stress σ 192 × 106N·m−2 Where, T is the wave period, g is the acceleration of gravity, h is water
depth.
For the wave force and current force on per unit length of a small-
4. Design load analysis scale component, Morrison formulate is used to calculate the drag force
and inertia force. And then two kind of forces are synthesized in the
Offshore platforms have long engaged in production operations in same phase, as shown in Eq. (10). For the platform rod, the pressure in
the sea. The sea and meteorological conditions in the working zone the wave load zone is shown in Eq. (11).
have a great impact on the structural safety and operational efficiency.
1
Therefore, determining the environmental load and weather conditions F= CD A |u x | u x + CM Vux
2 (10)
of the platform is the first task to be considered in the design of offshore
structures. Offshore structures bear a lot of environmental loads, in- 1
P= CD |u x | u x + CM u x
cluding wind load, wave load, current load, sea ice load and earthquake 2 (11)
load, as shown in Fig. 3. By referring to the relevant data, the hydro-
Where, A is the projection area per unit length pile column perpendi-
logical conditions of Liaodong Bay in the Bohai Sea are shown in
cular to the vector u x , V is the drainage volume for unit length of the
Table 3.
component. ux and u x are respectively the relative velocity and accel-
eration vectors perpendicular to the axial component, which obtained
4.1. Wind load by Stokes solution. CM is the inertia force coefficient, the size often
takes 1.6 [32]. Other symbols have the same meaning as before.
The wind drag force on an object should be calculated as shown in
Eq. (6) [31]. For the platform rod, the pressure in the wave load zone is 4.4. Ice load
shown in Eq. (7).
The force of sea ice on marine structure is called ice load. For off-
F = ( /2) u2Cs A (6) shore platform, cone ice breaking devices are often installed on the legs

41
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 3. Loads on offshore platform jacket structure.

Table 3
Environmental parameters conditions [30].
Item Conditions Parameters and values

Sea wind Operating conditions One minute operating status (once a year) is 30.6 m/s
Ultimate conditions Storm self-saving state (once in a century) is 46.31 m/s
Wave Operating conditions The maximum wave height is 1.4 m and period is 5s
Ultimate conditions The maximum wave height is 8.0 m and period is 9.6s
Current Operating conditions The maximum flow rate is 1.03 m/s
Ultimate conditions The maximum flow rate is 1.67 m/s
Sea ice General ice Year Ice thickness 15 cm; compression strength 1.5 MPa, bending strength 0.7MPa
Extreme Ice Age Ice thickness 40 cm, compressive strength 1.5 MPa, bending strength 0.7Mpa; Flow ice size 5 × 5×0.7m3; flow ice Speed 1.03 m/s

Table 4 Table 6
Wind Shape Coefficients. Environmental loads summary.
Area Shape Coefficient (CS) Type of load Wind Flowing ice Ice field

Beams 1.5 Load size(kPa) 1.31 3.534 480


Sides of buildings 1.5
Cylindrical sections 0.5
Overall projected wind area of platform 1.0
in this paper.
Two kinds of ice loads that mainly generate damages to marine
engineering structures are as following [32].

(1) The load of ice field on vertical pile. The ice load generated by
massive ice field squeezing isolated vertical piles is shown in Eq.
(12):

P = mK1 K2 Rc bh (12)

Where, m is the shape factor of pile column, see Table 5, K1 is the


local extrusion coefficient, K2 is the contact coefficient between the
Fig. 4. Schematic diagram of cone breaking ice structure. pile column and the ice layer, Rc is the ultimate compressive
strength of ice cubes, b is the width or diameter of piles, and h is the
thickness of ice.
Table 5
Shape coefficient of pile column. (2) The load of flowing ice on vertical pile. When the sea ice is carried
by wind and current, the maximum value of the ice pressure on pile
Tip Angle of pile tip(2α) 180° 120° 90° 75° 60° half round
column can be calculated by the Eq. (12). If the kinetic energy of ice
Coefficient (m) 1.00 0.81 0.73 0.69 0.65 0.90 can only partly cut into the pile, the ice will stay in front of the pile.
In this case, the maximum impact ice pressure acting on the pile
will be less than the above, but the impact of ice on the platform
near the sea surface to destroy the ice through the direct action of tri- should not be ignored. Therefore, the kinetic energy of flowing ice
angular breaking devices on the ice, so as to reduce the squeezing force is mainly consumed on the work of ice field squeezing. The ice
of the ice and reduce the load on the jacket platform. The principle of pressure P on the pile at any moment is shown in Eq. (13).
ice breaking device is shown in Fig. 4 and the ice breaking angle α is 60°
P = 0.68vh (BL Rm tan )1/2 (13)

42
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 5. a) The variation of wave and current load with water depth and time, b) wave and current load varies with water depth at 3/4 period.

Fig. 6. Flowchart of Topology optimization solution.

Where, L is the length of flowing ice, B is the width of flowing ice, h water depth and time. It can be expressed through the wave theory and
is height of flowing ice, and v is the velocity of flowing ice. Morrison equation, as shown in Fig. 5. It can be seen that the wave and
current load changes periodically, and have the maximum value at 3/4
period. Fig. 5 (b) indicates the load curve at 3/4 period. In the sub-
4.5. Design loads summary
sequent analysis, the maximum current and wave load value is applied
on the model. So the applied load is according to the load curve as
In this paper, the topological optimization analysis of continuous
shown in Fig. 5 (b).
body structure is carried out with wind load, current load and wave
load in extreme condition. The static analysis and dynamic analysis of
the jacket before and after optimization were carried out under the 5. Topology optimization design
same environmental loads.
Under various environmental conditions, load magnitude varies In this paper, the topology optimization method is used to ignore
little with height for wind load and sea ice load, so it is calculated as a the initial design of the structure completely, and the design space is
constant value. The calculated results are shown in Table 6. For the constructed as a monolithic continuum structure. The optimal response
wind load and current load, they could be calculated by the corre- and the target are set by adding reasonable constraints such as external
sponding criterion. For the ice load, the ice filed load is very large. loads and manufacturing, and the final structure is obtained through an
However, it is not considered in the following analysis, because the iterative solution. The goal of optimization is to increase the service life
jacket is often equipped with a cone ice breaking structure. of the structure and reduce the design and manufacturing costs. The
For the wave and current load, its magnitude varies a lot with the topology optimization process is shown in Fig. 6.

43
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 7. Designed space of topology optimization for jacket platform structure.

Fig. 8. Loads distribution on the topology optimization design space.

Fig. 9. Evolution in topology optimization of jacket structure.

44
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 10. Model reconstruction process under different filtration densities.

Table 7 other constraints.


Platform weight of different models. The initial structure can be optimized by different optimization
Pre-optimization Optimization model
methods and solution methods by a series of iterations. Until the opti-
mization iteration is convergence, the topology optimized results would
Model a Model b Model c be reconstructed and compared with the pre-optimization model to
evaluate its performance. If topology optimization cannot meet the
Weight of platform (t) 329.97 268.62 305.46 316.87
18.59% 7.43% 3.98%
expectation results, the design parameters need to be modified, and
structure would be re-topology optimized until a satisfactory topology
structure would be obtained. It is noted that the structure obtained by
For the initial design space, it is often determined according to the the topology optimization is not necessarily the final design. The opti-
shape of the initial model structure, and then use CAD software to es- mized model only reflects the force transmission path of the structure
tablish the continuum structure of the initial model and import CAE under the action of the external load. Therefore, it is needed to re-
software. Due to the different modeling mechanisms of CAD and CAE construction of the optimized model for the final structure.
software, the imported model needs to be cleaned up firstly before
analyzing imported models, such as removing shared edges, repairing 5.1. Modeling of jacket structure
missing faces and so on. Next, the imported model is meshed, and
added material properties and boundary conditions to perform finite Firstly, the optimal design area of jacket structure is determined.
element analysis on the model. And then the optimal response are set Considering the requirements of manufacturing, installation and other
on the initial model, such as compliance, frequency, volume, etc. In practical requirements, large design area is selected as much as possible
addition, the optimization still needs to consider the feasibility of to fully enlarge the optimization potential. For the leg of jacket plat-
processing and manufacturing constraints, such as draft, extrusion and form, it is often designed as the double diagonal state, which can

Fig. 11. Stress cloud of the jacket model before and after optimization.

45
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 12. Displacement cloud of the jacket platform before and after optimization.

Table 8
Static characteristics of the jacket platform before and after optimization.
Optimized project Pre-optimization Optimized Reduced amount Reduced ratio

Jacket weight (t) 329.97 316.87 13.1 3.98%


Maximum displacement(mm) 16.56 9.214 7.346 44.36%
Maximum stress(MPa) 59.6 22.4 37.2 62.4%

enhance the anti-overturning ability of platform structure. From the applied at the bottom and four symmetry constrains are applied at the
planar perspective, it can be regarded as a trapezoid structure with middle. Considering the actual interaction between pile leg of jacket
small top. and soil, pile structure in mud is generally simplified to the length of 6
Referring to the structure of JZ20-2MUQ jacket platform, only the times pile diameter, and fixed constraints are applied at the bottom.
underwater structure is selected to establish the three-dimensional Considering the external loads are not symmetric, two vertical ortho-
continuum model as shown in Fig. 7. To be consistent with the proto- gonal planes of symmetry need to be imposed as additional design
type model, the optimum model still adopts four legs supporting on the conditions to the optimum model.
lower and retains the cross braces on the upper. It can be clearly seen In order to obtain a relatively clear truss structure, the minimum
that the grey part is defined as the optimization space, while legs and flexibility of the model is taken as the optimization objective function,
braces are defined as non-optimization space and do not participate in and the volume fraction and maximum displacement of the model are
the optimization process but contribute to the overall flexibility. In restrained.
addition, in order to ensure the optimization result with a clear truss
structure, the internal material is removed during establishing the op-
timization model. And two layers of the solid structure are added at the 5.2. Results of topology optimization
middle position of original model, as shown in Fig. 7.
Details and distribution of loads on the optimization model is shown During the topology optimization process, the element density value
in Fig. 8. The design loads for the optimization are obtained from the of design space should be close to the ends of 0–1, so as to inhibit the
load analysis in Section 4, mainly including wind load, current load, generation of intermediate density and make the optimized structure
and wave load. The current and wave load both act on the optimization clearer. At the same time, it can be guaranteed that the optimized
space, but wind load partly act on the upper marine equipment and structure can reduce the structural weight in accordance with the safety
partly act on the optimization space. Therefore, wind load on the upper performance, so as to obtain better distribution of materials.
marine equipment is acted onto the cross brace above the optimization Topology optimization of the model was carried out using the
zone with additional torque. In addition, the gravities of upper marine OptiStruct optimization module in HyperWorks software [33]. The
equipment and lower section are both considered. For marine equip- optimum objective function is the minimization of compliance for de-
ment on the upper section of jacket, the equipment weight is equivalent signed structure, maximizing the stiffness, while reducing structural
to four mass points and act on legs. volume of the design-space. The process of topology optimization is
The size of load is changing with the depth, so the effect of load on shown in Fig. 9. After 73 iterations, the optimization process converges
the model can obtained by interpolation algorithm. Then, force equa- and a clear topological truss structure is obtained. Fig. 9 shows the
tions are generated in OptiStruct software, which are applied on the evolution in topology of the jacket structure, together with the values of
model. Compared with the operating conditions, the platform has a volume and relative stiffness for each iteration. It is noted that how the
large static stress and displacement under the conditions of extreme elements with low sensitivity value are being gradually eliminated,
marine environment. Therefore, in the topology optimization process, decreasing the volume of the piece until the conditions are satisfied and
the environmental load of the extreme condition is often used as the the solution of maximum stiffness is found.
control load of the platform design. For the last topology iteration results in Fig. 9, two layers of X-
For the model boundary constraints, four fixed constrains are shaped structure are appeared in the designed zone, which is in line
with the design principles of marine engineering structures. However,

46
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 13. Vibration modals of platforms before and after optimization.

Table 9 0.416, one layer structure emerges inside of structure. And when the
Modal frequency of platform before and after optimization. filtration density is 0.328, two layer structure emerges. As the de-
Modes Pre-optimization Optimized
creasing of filtration density, less materials are removed.
The final model could be reconstructed by the optimization results.
1st mode 1.7320 3.1171 For different optimized truss structures, cylindrical pipe material with a
2st mode 1.7423 3.1823 certain thickness is chosen to obtain the platform structure as shown in
3st mode 1.9963 3.1827
4st mode 2.4499 3.2273
Fig. 10. The pipe construction can better withstand the axial force,
5st mode 2.4645 3.3159 bending moment, torque and hydrostatic pressure, is widely used in
6st mode 3.0236 3.3178 marine engineering buildings. For the weight of platform, the three
optimization models have reduced to some extent. Especially, model a
has reduced the most weight to 18.59%, as it has the simplest structure.
optimized structure have no internal structures. It is because that the According to the topological optimization results, the three models
element sensitivity filtering value is set to 0.515, the elements with a, b, and c all can be used as the initial structure of further detailed
lower sensitivity values inside structure are removed in the process of design. Considering that the traditional jacket structure is often con-
optimizing iteration. It also shows that the outer structure is the main nected by the horizontal supports, model b and c both meet the re-
force transmission path. But compared to the classical model obtained quirements. But model c has more horizontal poles in the lower part.
by the conventional trial and error design method, the optimized Model b is selected for the final initial structure of jacket platform. The
structure without inner structure may cause unstable of the overall optimization process itself ends with the transformation of the shape-
platform. To get the inner structure, element sensitivity filtering value optimized data into a geometry-based CAD model, which itself has to be
should be reduced gradually until the optimization result of the inner meshed and reanalyzed using FEA in order to validate the final design.
structure appears, as shown in Fig. 10. When the element sensitivity is For ensuing the consistency of the analysis before and after optimized

47
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

Fig. 14. Time-displacement curve of platform top points (a) pre-optimized model, (b) optimized model.

Fig. 15. Time-stress curve of platform top points (a) pre-optimized model, (b) optimized model.

6. Discussion and verification

The optimum solution is next verified by FE analysis and its per-


formance is compared to the initial design under the same conditions. A
clear truss structure was obtained by topology optimization of the
model optimization area, and the model was reconstructed using
PIPE59 element in ANSYS. Platform models before and after optimi-
zation are shown in Fig. 10. The final weight of the optimized model is
7.43% lighter than the classical model. It can be seen that the new
model increases an X-shaped structure on the upper part, reduces the
horizontal braces structures and enlarges the X-shaped structure of
lower part simultaneously. Based on the previous research, the static
Fig. 16. Relationship of external load and displacement of jacket structure. performance, dynamic performance and ultimate carrying capacity of
platforms before and after optimization in the real marine environment
parameters are compared (Table 7).
Table 10
Ultimate carrying capacity of pre-optimization and optimized jacket platform.
6.1. Static analysis of ultimate state design
UCC (KN) 0° direction 45° direction
Stress distribution within such a large structure is a dominant factor
Pre-optimized 2820 2660
Optimized 5680 4400 in the design procedure of an offshore structure. Ultimate state analysis
Increased ratio 101.42% 65.41% is performed for the jacket platform under the ultimate environmental
load applied on the FEM model, which is list in Table 3. The equivalent
stress distributions and the displacement cloud before and after opti-
model, the upper frame structure of jacket platform is also added to the mization is shown in Figs.11 and 12.
optimized model, and pipe sizes of optimized model is the same as that It can be seen from Figs. 11 and 12 that the stress and displacement
of the original model. of the optimized model are greatly reduced. Compared with the pre-
It can be concluded from the above optimization results that the optimization model, the stress distribution of the optimized model is
topological optimization method is a kind of conceptual design method, more uniform. And its maximum equivalent stress value is only
which only provides the optimal transmission path. Further optimiza- 27.16 MPa, which is reduced by 46.31% comparing to the pre-optimi-
tion is needed for the specific size and relative position of structures. In zation model. It is proved that the optimization structure has a good
order to get better structural performance, detailed shape and size op- force transmission path and is better to withstand the environment load
timization are necessary after topology optimization. with less materials. In Fig. 12, main legs of optimized model have no

48
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

deformation, and the maximum displacement value appears on one X of the jacket structure is obtained. It can be clearly seen that the opti-
node. For the pre-optimization model, the main leg deformations are mized structure have more UCC under the same deformation dis-
very obvious, and large deformations are appearing on three X-shape placement.
structures with large displacement values. The static characteristics are UCC of the pre-optimized and optimized model can be obtained, as
listed in Table 8 in detail. shown in the Table 10. It can be seen that the ultimate carrying capacity
of the optimized structure has increased 101.42% and 65.41% for 0°
6.2. Dynamic analysis of ultimate state design and 45° direction respectively. Therefore, the optimized structure not
only has lighter weight but also has better ultimate carrying capacity,
Because the jacket platform is very susceptible to wind, current and which can improve the safety of the whole structure.
wave loads during their serve life, it is very important to evaluate its
dynamic response for the platform model before and after optimization. 7. Conclusion
The natural periods and corresponding vibration mode shapes are
computed by analysis. The first six dominant vibration modes are This paper presents the topology optimization method in the design
shown in Fig. 13 and the natural period Modal frequency are listed in process of offshore jacket platform. Topology optimization offers new
Table 9. potential in the industrial design process, which results in lighter,
Comparing to the pre-optimization model, the modal frequency of stronger, and more durable structures. The underwater jacket structure
each order for optimized model are all bigger, which is far from the is selected to determine the design area. A clear truss structure is ob-
resonant frequency under wave loads. Additionally, the first six modal tained through a series of topology iterations. The optimization results
frequency optimized structure have little differences. The maximum are verified by FE analysis for the static performance, dynamic per-
deformation values of various order are much bigger, and it is located at formance, and ultimate carrying capacity. The following conclusions
the X-shape of lower part. The maximum deformation of pre-optimi- can be drawn from this study:
zation model is located at the upper part. There are two braces layers
for the pre-optimization model, which can resist the deformations of (1) The topological optimization method based on SIMP have good
middle structures. But the bigger deformation of optimized model has effect for the jacket structure design.
much influences on the dynamic characters of platform. Based on the (2) By using the proposed optimization model, the mass of the jacket
static analysis, the modal deformations are still smaller than the ulti- structure is reduced by 13.7%. This indicates the mass of the jacket
mate deformations. Also, the modal frequencies are far from the re- structure can be significantly reduced by using the proposed opti-
sonant frequency, which reduces the resonance probabilities. mization model.
The transient dynamic analysis is also compared before and after (3) The static analysis results indicate that the maximum equivalent
optimization. The maximum displacement and stress over time are stress value is reduced by 46.31%. It is proved that the optimized
extracted, which is located at the top and bottom part, respectively. The jacket structure has a better force transmission path and is better to
maximum displacement changing curves over time are shown in withstand the environment load with less materials.
Fig. 14. Platforms would sway under the effect of environmental loads. (4) The dynamic analysis results show that the modal frequencies are
The maximum displacement amplitude of pre-optimization platform is far from the resonant frequency, which reduces the resonance
15 mm, but the maximum displacement amplitude of optimized model probabilities.
is only 2 mm. It can be seen that, the displacement of optimized model (5) The limit analysis results show that UCC of optimized platform
has decreased a lot, which increase the stability largely. The stress structures are both higher than the pre-optimized structures whe-
curve of before and after optimized model are shown in Fig. 15. It can ther it is 0° or 45° load direction.
be seen that the stress amplitude is reduced from 15.1Mpa to 2.2Mpa,
which reduced the stress sharply. It can be conclude that the optimized These positive results indicates that topology optimization is a
model has good performance under the dynamic loads. powerful tool to find the optimal material distribution for the pre-de-
sign of engineering structures. It may be a perspective design method to
6.3. Limit analysis other offshore structures in the future.

The linear static analysis of the optimization results can only give Acknowledgments
the laws of stress, strain and displacement of the material in the elastic
state. As the external load increases, the deformation of the jacket The research was supported by National Natural Science Foundation
structure will enter the plastic state, and the relationship between stress of China (51609223, 51606178), Projects of Qingdao Application Basic
and strain will change into a nonlinear relationship. When the external Research (16-5-1-21-jch), and National Science Fund for Distinguished
load reaches to a certain limit, the deformation will increase un- Young Scholars (51625902).
limitedly, resulting in the loss of carrying capacity of the structure. This
state is the plastic limit state of the structure, and the load at this time is References
the Ultimate Carrying Capacity (UCC) of the structure. Therefore, in
order to explore the UCC of the optimization results, it is necessary to [1] M. Jang, L.P. Li, Offshore Engineering Structures—Jacket Platform, Dalian
analyze the plastic deformation state of the jacket structure. University of Technology Press, Dalian, 2009.
[2] T. Taghikhany, S. Ariana, R. Mohammadzadeh, S. Babaei, The effect of semi-active
When performing the limit analysis, the finite element models of controller in sirri jacket seismic vibration control, Int. J. Mar. Sci. Eng 3 (2) (2013)
pre-optimization and optimized are both reestablished by BEAM 188 77–84.
element. The external environmental load is converted into a con- [3] K. Wei, S.R. Arwade, A.T. Myers, S. Hallowell, J.F. Hajjar, E.M. Hines, W. Pange,
Toward performance-based evaluation for offshore wind turbine jacket support
centrated force acting on the model. Considering the loading of 0° and structures, Renew Energ 97 (2016) 709–721.
45° direction, the limit analysis of platform was carried out by the in- [4] S. Feng, Y.P. Song, R.X. Zhang, Optimum design of structure shape for offshore
cremental method. With the increasing of the external environmental jacket platforms, China Ocean Eng 14 (4) (2000) 435–445.
[5] X. Liu, G. Li, Q.J. Yue, R. Oberliers, Acceleration-oriented design optimization of
load, the maximum displacement of structure is increasing and reaches ice-resistant jacket platforms in the Bohai Gulf, Ocean Eng 36 (2009) 1295–1302.
the maximum value under the ultimate load, as shown in Fig. 16. When [6] H.Z. Yang, Y. Zhu, Q.J. Lu, J. Zhang, Dynamic reliability based design optimization
the applied load exceeds the limit state, the deformation of the jacket of the tripod sub-structure of offshore wind turbines, Renew Energ 78 (2015)
16–25.
platform increases sharply, resulting in the calculation result no-con-
[7] Y.S. Lee, B.L. Choi, J.H. Lee, S.Y. Kim, S. Han, Reliability-based design optimization
verging, and the calculation would stop. At this point, the ultimate load

49
X. Tian et al. Applied Ocean Research 84 (2019) 38–50

of monopile transition piece for offshore wind turbine system, Renew Energ 71 [20] Michael Yu Wang, Xiaoming Wang, Dongming Guo, A level set method for struc-
(2014) 729–741. tural topology optimization, Comput. Methods Appl. Mech. Engrg 192 (2003)
[8] S. Chen, G. Fu, A fuzzy approach to the lectotype optimization of offshore plat- 227–246.
forms, Ocean Eng 30 (2003) 877–891. [21] H.A. Eschenauer, N. Olhoff, Topology optimization of continuum structures: a re-
[9] K.H. Chew, K. Tai, E.Y.K. Ng, M. Muskulus, Optimization of offshore wind turbine view, Appl. Mech. Rev. 54 (2001) 331–390.
support structures using an analytical gradient-based method, Energy Procedia 80 [22] Liu Yuhui, Study on Optimization Design of Jack-up Platform With Three Legs,
(2015) 100–107. Master Thesis China University of Petroleum (East China), 2013.
[10] T. Gentils, L. Wang, A. Kolios, Integrated structural optimisation of offshore wind [23] Lu Xingguo, Structural Strength Analysis and Optimum Design of Pile Leg of Self-
turbine support structures based on finite element analysis and genetic algorithm, elevating Offshore Platform, Master Thesis Zhejiang Ocean University, 2016.
Acs Appl. Energy Mater. 199 (2017) 187–204. [24] A.G.M. Michell, The limits of economy of material in fame structure, Philos. Mag.
[11] B. Torstenfelt, A. Klarbring, Conceptual optimal design of modular car product Abingdon (Abingdon) 8 (6) (1904) 589–597.
families using simultaneous size, shape and topology optimization, Finite Elem. [25] K.M. Zhou, J.F. Li, X. Li, A review on topology optimization of structures, Adv Mech
Anal. Des. 43 (14) (2007) 1050–1061. 35 (1) (2005) 66–76.
[12] R.Z. Gustavo, J.N. Antonio, L. Francisco, A.C. Carlos, A survey of multi-objective [26] K. Svanberg, The method of moving asymptotes: a new method for structural op-
metaheuristics applied to structural optimization, Struct. Multidiscipl. Optim. 49 timization, Int. J. Numer. Anal. Methods Eng. 24 (1987) 359–373.
(2014) 537–558. [27] O. Sigmund, J. Petersson, Numerical instabilities in topology optimization: a survey
[13] X. Yan, X. Huang, Y. Zha, Y.M. Xie, Concurrent topology optimization of structures on procedures dealing with checkerboards, mesh-dependencies and local minima,
and their composite microstructures, Comput. Struct. 133 (2014) 103–110. Struct. Multidiscipl. Optim. 16 (1998) 68–75.
[14] Y.Q. Wang, Z. Kang, Q.Z. He, An adaptive refinement approach for topology opti- [28] Offshore Oil Engineering Design Guide, Offshore Oil Engineering Platform Structure
mization based on separated density field description, Comput. Struct. 117 (2013) Design Vol. 4 Petroleum Industry Press, Beijing, 2007.
10–22. [29] Q.Z. Yan, F.J. Yang, The reliability analysis of offshore platform under environ-
[15] L.B. Lauren, B. Alessandro, K. Neil, F.B. William, H.P. Glaucio, Connecting archi- mental load in Bohai bay, Adv. Sci. Technol. Eng. Syst. J. 11 (24) (2011) 5782-07.
tecture and engineering through structural topology optimization, Eng. Struct. 59 [30] API R, Recommended Practice for Planning, Designing, and Constructing Fixed
(2014) 716–726. Offshore Platforms—working Stress Design, 22st ed., API Publishing Services,
[16] J.H. Zhu, W.H. Zhang, L. Xia, Topology optimization in aircraft and aerospace Washington, D.C, 2014 API RP 2A-WSD.
structures design, Arch. Comput. Methods Eng. 23 (2016) 595–622. [31] D.C. Sun, B. Pan, Design and Research of Ocean Jack-up Mobile Platform, Shanghai
[17] G.I.N. Rozvany, A critical review of established methods of structural topology Jiao Tong University Press, Shanghai, 2008.
optimization, Struct. Multidiscipl. Optim. 37 (2009) 217–237. [32] M.W. Dingemans, Water Wave Propagation Over Uneven Bottoms, Delft University
[18] M.P. Bendsøe, N. Kikuchi, Generating optimal topology in structural design using a of Technology, TU Delft, 1994.
homogenization method, Comput. Methods Appl. Mech. Eng. 71 (1988) 197–224. [33] Q.Q. Hong, K. Zhao, Zhang P. OptiStruct, HyperStudy: Theoretical Foundation and
[19] Fei Zhao, A nodal variable ESO (BESO) method for structural topology optimiza- Engineering Application, China Machine Press, Beijing, 2012.
tion, Finite Elem. Anal. Des. 86 (2014) 34–40.

50

View publication stats

You might also like