0% found this document useful (0 votes)
13 views61 pages

Chap_4___Computer_Aided_Design_Optimization

This document discusses the importance of optimization in the structural design of tall buildings, highlighting the transition from traditional trial-and-error methods to modern numerical optimization techniques. It outlines the design process, including establishing lateral stiffness criteria and selecting material strengths, while emphasizing the advantages of optimization such as reduced design time and improved quality. Additionally, it addresses the limitations of optimization methods and presents examples of design problems and solutions.

Uploaded by

Derek Lam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views61 pages

Chap_4___Computer_Aided_Design_Optimization

This document discusses the importance of optimization in the structural design of tall buildings, highlighting the transition from traditional trial-and-error methods to modern numerical optimization techniques. It outlines the design process, including establishing lateral stiffness criteria and selecting material strengths, while emphasizing the advantages of optimization such as reduced design time and improved quality. Additionally, it addresses the limitations of optimization methods and presents examples of design problems and solutions.

Uploaded by

Derek Lam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 61

- 4.

1 -

COMPUTER AIDED DESIGN OPTIMIZATION


Introduction

 The concept of optimization is basic to much of what we do in our daily


lives. We would always like to do or be the best in some sense. For
example, the desire to excel, to find the shortest path to reach a
destination, to get the highest mark in examination.

 In structural engineering of tall buildings, the design practice is


traditionally a very time consuming and laborious task.

 Although today’s finite element computations allow for precise analysis


of large-scale tall building structures, they are still lacking for the ability
of providing insight for economical design.

 Unlike structural analysis, structural design is a decision making process.


Conventionally, the structural synthesis process is carried out by
engineers based on their experience, intuition and repetitive trial-and-
error.

 Today, the emerging structural optimization technology provides a


promising design tool to automate the structural synthesis process and
to aid in searching for the best design to meet various design
requirements. Indeed, the structural optimization provides a new design
paradigm. It offers engineers a systematic and rational approach to
design decisions where we traditionally rely heavily on subjective
intuition and experiences.

 However, this should not be construed to suggest that the optimization


design process can be reduced to a few computer runs or that our
intuition and experience are unimportant. Rather, the numerical
optimization can now be used to relieve us of the tedious repetitive
design calculations, freeing us to spend time on the truly creative aspects
of engineering design.

 Motivation of this section of the course is to introduce the concepts of


numerical optimization and to provide a solid understanding of the
emerging optimization techniques for the design of practical tall building
structures.
- 4.2 -
Tall Building Design

 A primary concern for the structural design of a tall building is to


adequately control its lateral movement induced by wind loads.

 In fact, one dividing line between tall buildings and other multistorey
buildings is precisely at the point where design for motion under lateral
loads becomes more critical than design for strength.

 Tall building design process

1) Establish lateral stiffness criteria.

2) Define the lateral-load resisting system which results in the least


material to provide sufficient level of stiffness.

3) Finally, choose the strength level for the material to safely sustain all
stresses.

 Traditionally, much attention has been absorbed in the design for strength
(e.g. code-stipulated requirements), which is however the less important
part of the structural design of a tall building.

 Creation of a lateral-load resisting system for a tall building is the most


challenging task for a structural engineer. Once the topology of the
structural system for a tall building is defined, the main effort is to size the
structural members to satisfy all performance criteria.

 Economically sizing members of a tall building to satisfy drift requirements


is generally rather difficult since drift constraints are system level
constraints which require simultaneous consideration of all members
entailed in the structural system.
- 4.3 -
Conventional vs Numerical Optimal Design Methods

 Conventional Design Method

The traditional design method is an educated trial-and-error process.


Once the design problem is proposed, efforts are made to estimate the
initial design, compute the responses and manually make design
changes to improve efficiency whilst complying with various design
performance criteria.
- 4.4 -
Common criticisms of traditional design approach:

- No guarantee for optimality


- Design improvement by subjective intuition and experiences
- Time consuming process
- Design aimed to simply satisfy code requirements and regulations
- Lack of rigorous cost evaluation on design alternatives

 Numerical Optimal Design Method

Modern numerical structural optimization methods are goal oriented


synthesis techniques that optimise structures for the purpose of
achieving the best value at the lowest overall cost while satisfying all
specified design criteria. This approach requires the mathematical
formulation of the design problem, in which the design objectives, the
design constraints and the design variables must be explicitly identified
and mathematically defined.

Correct formulation of a design optimization problem is important and


generally uses the following steps.

- Identification and definition of independent design variables.

As a first step in the problem formulation phase, the variables that


describe the design of the structural system must be identified and
defined. It is necessary that all the design parameters and data related
to the problem must be precisely defined. These include material
properties, cross-sectional properties of members and the like.

- Identification and definition of an objective function.

As a second step in the problem formulation phase, an objective function


(also known as the cost or merit function) that measures the relative merit
of alternate designs needs to be identified and defined. This is a scalar
function that must be expressed in terms of the design variables.
Common objective functions are cost of the system, weight of the
structure, maximum deflection, natural frequency, and so on.
- Identification and definition of design constraints
- 4.5 -

Each design problem has constraints that must be met, such as lateral
drift and code-specified element strength design requirements. Design
constraints are generally related to the limitations on the structural
response performance as well as explicit size bounds on the design
variables. Typical examples of the structural behavioural responses to
be constrained are stresses, lateral deflections and natural frequencies.
- 4.6 -
Advantages and Limitations of Structural Optimization

 Advantages:

- A major advantage is the reduction in design time.

- Design automation can be achieved with quick turnaround time and


quantitative feedback.

- Design is not based by intuition and experience in engineering.

- Optimization provides a systematic logical design procedure.

- Able to deal with a large number and a wide variety of design


variables and constraints.

- Optimization virtually always improves the quality of design if the


design problem formulation is well posed.

 Limitations

- Incomplete problem formulation may lead to meaningless if not


dangerous design.

- Optimization techniques are limited to the range of applicability of the


analysis model.

- Computational effort generally increases as the number of design


variables increases.

- Ultimate global optimum is not always guaranteed.

- Cannot handle design problems that cannot be explicitly defined.


- 4.7 -
Example: Design of a 2-bar truss to minimise the total volume of material
while limiting the top deflection to be within 0.1.

Nondimensional data

E=100, L1=20, L2=10

Initial sizes: A1=A2=1.0

Design Problem Formulation:


- 4.8 -
Design Solution:

Top displ = 0.1

Feasible
Region

Vol = 900

Vol = 833.3
- 4.9 -
Optimal Stiffness Design Technique

Intuitive Method

 Practising engineers involved in the design of tall buildings have long


concerned with the stiffness optimization of tall buildings and have
developed computer software for sizing members of tall building
frameworks subject to displacement constraints.

 Examples of some notable professionals who have published their


stiffness optimization methods:

- C.H. Thornton et al (1990)1, Thornton-Tomasetti Engineers, New


York
- W. Baker (1990)2, Skimore, Owings & Merill, Chicago
- F.A. Finley (1991)3, Gilsanz and Carlson (1993)4

 Their method is all essentially based on an intuitive concept of uniform


strain energy density and is applicable only for single displacement
constrained problems. The optimization algorithm is listed as follows:

1) The method first calculates the contribution of each member to a


particular deflection of concern. Using the principle of virtual work, the
contribution of each member to resist the deflection of interest is
indeed the virtual strain energy of each member.

1
C.H. Thorton, L. Joseph and T. Scarangello, “Optimization of Tall Structures for Wind
Loading”, J. of Wind Engineering and Industrial Aerodynamics, 36, 1990, 235-244.
2
W. Baker, “Sizing Techniques for Lateral Systems in Multi-story Steel Buildings”, Proc.
Fourth World Congress – Tall Buildings: 2000 and Beyond, CTBUH, Hong Kong, Nov. 5-9,
1990, 875-868.
3
F.A. Charney, “The Use of Displacement Participation Factors in the Optimization of Wind
Sensitive Buildings”, Proc. Ninth Structures Congress, ASCE, Indianapolis, Indiana, Apr. 29
– May 1, 1991, 545-547.
4
R. Gilsanz and A. Carlson, “Optimization in Building Design”, Proc. Second Int. Conf. on
Computer Aided Optimum Design of Structures 91, Boston, 1991, 541-551.
- 4.10 -
Let’s consider the tip deflection  of a building of N members.

If the building is a truss,

N Li
    
FX f X 
 dx
i 1 0  EA  i

2) To achieve the optimal design by enforcing every structural member


of a framework to contribute the same amount of work per unit volume
of material to resist the occurrence of the specific top deflection of
concern to the design.

 FX f X L 
If the virtual work of a truss element = Wi    , then the work
 EA i
 F f L W
density of the element = Di   X X   Ai Li  i .
 EA i Ai Li

 W   W   W 
At optimum,         ...  constant .
 A L 1  A L 2  A L 3

3) For statically indeterminate structures, one needs to update member


sizes, reanalyse the structure and repeat the optimization routine until
convergence to the optimal design is obtained.

 Although the intuitive method is quite efficient and easy to be coded, it is


useful only for single displacement constrained problems of statically
determinate truss structures.
- 4.11 -
Example of the 2-bar truss,
- 4.12 -
Formal Optimization Method

 As modern tall buildings are complex structures which often require the
consideration of multiple drift constraints under multiple lateral loading
conditions, it is desirable to have an efficient resizing technique for the
optimal design of tall buildings subject to multiple constraints.

 A rigorously derived Optimization method for the stiffness design of tall


building frameworks has been developed and is herein briefly presented.
Further details can be found in Chan (1997)5.

 Consider a building structure having N members, M number of drift


constraints and P element strength constraints. The design optimization
problem of finding minimum structure weight/cost W can be posed
implicitly as
N
Minimize W ( Ai )   wi Ai
i 1

subject to  j   Uj ( j  1,2,..., M )
 p   Up ( p  1,2,..., P )
AiL  Ai  AiU (i  1,2,..., N )

where wi = weight/cost coefficient; Ai = sizing variable


 j = jth drift constraint;  Uj = corresponding drift limit
 p = pth stress state;  Up = corresponding strength limit
AiL , AiU = lower and upper size bounds

 Before solving the above optimization problem, one needs to first


express explicitly the drift constraints in terms of design variables. At this
end, let’s consider the design of tall steel buildings and the design of tall
reinforced concrete buildings separately.

5
C.-M. Chan, “How to Optimize Tall Steel Building Frameworks”, In: ASCE Manuals and Report on
Engineering Practice No. 90 - Guide to Structural Optimization, ASCE, 1997, 165-196.
- 4.13 -

Tall Steel Building

 For the building is a general 3D framework

N Li  FX f X FY fY FZ fZ M X mX MY mY MZ mZ 
         dx
i 1 0  EA GA Y GAZ GI X EI Y EI Z i
axial shear torsion bending

where FX, FY, FZ, MX, MY, MZ = actual member forces & moments
fX, fY, fZ, mX, mY, mZ = virtual member forces & moments
A, AY, AZ, IX, IY, IZ = sectional properties
E, G = material properties

 As practical steel buildings require the use of commercial standard steel


sections, the cross-section properties AY , AZ , I X , IY , I Z may be all
instantaneously expressed in terms of the axial area A using the
following regressional relationships

1 C 1 C
 AY  C AY
'
;  AZ  C AZ
'
;
AY A AZ A
1 C 1 C IY
 IX  C IX
'
;   C IY
'
;
IX A IY A
1 C
 IZ  C IZ
'

IZ A

where coefficients C and C’ are determined by regression analysis

 Substituting the regressional relationships into the virtual work


formulation, one obtains

N  eij 
 j ( Ai )     eij' 
i 1  Ai 
- 4.14 -
Li
F f  MY mY CIY  M Z m Z CIZ
where eij    X X 
0
 E
FY fY C AY  FZ f Z C AZ  M X m X C IX 
 dx
G i
Li  M m C'  M m C'
  
Y Y IY Z Z IZ
eij'
0 E
'
FY fY CAY  FZ f Z CAZ
'
 M X m X CIX
'

 dx
G i
- 4.15 -

 Note that the structural design of tall buildings is mostly controlled by


serviceability design criteria in term of lateral drift, the element strength
constraints can therefore be treated as secondary constraints.

 Recognizing this fact, one can get around the element strength
constraints by considering not to include them explicitly in the design
optimization process.

 Unlike drift constraints, which are system-level constraints, member


strength requirements are local-level constraints. Immediately after a
steelwork is analysed under different gravity and lateral loading
combinations, the element strength sizes can then be determined in
accordance with code requirements in a member-by-member manner.
Indeed, such strength based design procedures are now available in
most structural design software.

 Once the strength-based element sizes are obtained, they can be taken
as the lower-size bounds on the sizing variables in the lateral drift design
optimization.

 Upon formulating the design problem explicitly in terms of the design


variables Ai , the constrained design optimization problem can be
reformulated as an unconstrained minimization of the Lagrangian
function

N M  N  eij  
L( Ai ,  j )   wi Ai    j    eij'    Uj 
i 1 j 1  i1  Ai  

where the Lagrange multipliers  j must be non-negative such that

j 0 if  j   Uj (active constraints)

j  0 if  j   Uj (inactive constraints)
- 4.16 -
 Optimality Criteria (OC) Method:

There exists a number of optimization methods to solve the lateral drift


design problem of tall building structure. The most effective method
developed is based on a rigorously derived Optimality Criteria (OC)
method, which has been shown to be computationally very efficient for
large-scale structures.

The OC method involves two complementary phases:

Phase 1: Derivation of Optimality Criteria

- Differentiate L( Ai ,  j ) with respect to Ai , set the derivative to zero


and rearrange terms to obtain the Optimality Criteria

M eij
j 1 (i  1,2,..., N )
j 1 wi Ai2

Note that  j can be interpreted as a weighted factor which


measures the importance of the jth constraint to the optimal design.
The Optimality Criteria specify that the weighted sum of the virtual
strain energy densities for each member must be uniform and
indeed equal to unity.

Phase 2: Application of a recursive algorithm to achieve the optimum


by satisfying the Optimality Criteria

Recursive relation for Ai :

 1  M  j eij  
Ai1  Ai  1     1 ( i  1,2,...,N A )
   j 1wi Ai2  
 

where  ,  1 represent two successive iteration numbers;  is a


relaxation parameter which controls the rate of convergence and
N A is the number of active sizing variables having a member size
value to be within sizing limits.
- 4.17 -

Recursive relation for  j :


By considering the change ( Uk   k ) due to ( Ai 1  Ai ) to get a
system of M linear equations for M number of  j

NA  e e  NA
 
M
e
  
 j  ij ik3 = ik
 
   kU   k (k  1, 2,.., M )
j 1 i 1  wi Ai  i 1 Ai

The Gauss-Seidel iterative technique is adopted to solve for the


above simultaneous equations. Any  j value found to be negative
is adaptively set to zero to ensure for the non-negative sign
restriction for  j .

The above two recursive relations form the basis of an automatic OC


resizing method to solve for the stiffness optimization problem. Upon
convergence of the values for Ai and  j , a continuous-valued
optimal design solution is obtained

Initial Design Selection

 From the recursive resizing relations, it can be observed that the rate of
convergence of the OC process depends on the initial set of member
0
sizes Ai adopted at the beginning of the design process.

 In the following, an initial sizing preprocessor, which is based on the


“closed form” solution of the single-constraint problem, is presented to
proportion the initial design after the first analysis to establish a
reasonable starting design for the iterative OC process.

 If a single drift constraint is alone of concern, the optimality conditions


can be given as
eij
j 1 (i  1,2,..., N )
wi Ai2
- 4.18 -

Rearranging terms, one can obtain each section area Ai as

Ai   j  eij wi (i  1, 2,..., N )

To determine the as yet unknown Lagrange multiplier  j , one can set

N eij  N  eij  wi 
 j    
U
 eij    
'
 eij 
'
i 1    e w  i 1  j 
j
 j ij i   

and then solve for the value of Lagrange multiplier  j as


N
 eij  wi
j  i 1
N
   eij'
U
j
i 1

Finally, the initial member sizes can then be determined analytically


from the following “closed form” solution of the single-drift design
problem as

 N ew 
 i ij i  eij
Ai   1 N   (i  1,2,..., N )
  j   eij  wi
U '
 i 1 

Pseudo-Discrete Section Selection

 At the convergence of the OC iterative procedure where the optimum


 1 
values of element sizes are determined ( Ai  Ai ) and the active
 1 
constraints are identified (i  i ) , the solution normally results in
the continuous-valued member sizes.
- 4.19 -

 A common approach to finalise member sizes is to round up all the


continuous member sizes to their next available standard steel section
sizes. However, while this simple round-up technique is direct, it
produces a more conservative design than necessary.

 An effective pseudo-discrete section selection strategy is developed in


the following to achieve a smooth progressive transition from the
continuous-variable design to the final optimum design using discrete
standard steel section (Chan 19926).

 To minimise the least deviation from the deviation from the continuous
sizes to the discrete sizes, one effective strategy is to select the member
involving the least change in material weight when the continuous size is
increased to the next available discrete size, i.e., the member with the
least value of wi  Ai ( discrete )  Ai ( continuous )  .

 Once the currently most economical member is assigned a standard


section, its discrete size is then fixed to become inactive in the
subsequently section selection process. Since such member has been
assigned a discrete size which is larger than necessary, the yet
unassigned members may then be reduced in size. To do this, we can
update the value of the Lagrange multipliers progressively as

 eij eik  N  eik


N 
 
M
 j   3 
= 
   kU   k (k  1,2,.., M )
j 1 i 1  wi At  i 1 Ai

where  = number of inactive members currently assigned standard


sections.

6
C.-M. Chan, "An Optimality Criteria Algorithm for Tall Steel Building Design Using Commercial Standard
Sections", Structural Optimization, Vol. 5, 1992, 26-29.
- 4.20 -
Overall Design Optimization Procedure

j j

j
- 4.21 -
Examples: A series of Planar Steel Frameworks
- 4.22 -
Example: Three 50-story Frameworks
- 4.23 -
- 4.24 -
- 4.25 -
- 4.26 -
- 4.27 -
- 4.28 -
Tall Concrete Frame Building

 Unlike steel buildings which generally use standard steel sections for
members, tall reinforced buildings have more flexibility in forming their
member shapes and dimensions.

 Let’s assume reinforced concrete frame members are of rectangular


shapes.
MY
FY MX
L FX
MZ
FZ
B
Y
X D

 Taking the width (B) and the depth (D) of a member as design variables
for the member, the cross sectional properties of the member can be
expressed as
5
A  B  D, AY  AZ  B  D
6
1 1
IY  B 3  D , I Z  B  D3
12 12
1 B B4
I X    0.21 (1  )  D  B 3
3 D 12 D 4

Note that the effect of member torsion is generally small and assume
that B  D , the torsional moment of inertia I X can be approximated as
1
I X  D  B3
3

 Using the above sectional relationships and the principle of virtual work,
the drift constraint can be explicitly expressed in terms of design
variables B and D as follows:
- 4.29 -
N  e0ij e1ij e2ij 
 j ( Bi , Di )       j
U

i 1  Bi Di Bi Di Bi Di 
3 3

F f
Li
F f  FZ f Z 
where e0ij    X X  Y Y  dx
0
 E 5G / 6  i

M m 
Li
e1ij    Z Z  dx
0
 E / 12  i

M m M m 
Li
e2ij    Y Y  X X  dx

0 E / 12 G / 3 i

 Using the OC Method to first derive the Optimality Criteria and then apply
the following recursive algorithm to obtain the optimal solution

Recursive relations for Bi and Di respectively :

 1  M  j  e0 ij e1ij 3e2 ij   
B v 1
 B  1   
v
 2 2  2 4  4 2   1  (i  1,2,.., N B )
   j 1 wi  Bi Di Bi Di   v 
i i
Bi Di

 1  M  j  e0 ij 3e1ij e2 ij   
Div 1  Div  1     2 2   2
 1  (i  1,2,.., N D )
   j 1 i
w  Bi Di Bi
2
Di
4
Bi
4
D i   v 
where N B is the number of active width sizing variables and N D is the
number of active depth sizing variables.

Recursive relations for  j :

N B    v1 N D   
Considering  kU   kv    k
 i
i 1   Bi 
B  
Bi
v
 
  k  Di  Di ,
i 1   Di 

v 1 v
one

obtains a system of simultaneous linear equations for  j as


- 4.30 -

NB 1  e1ij 3e2 ij   e1ik 3e2 ik  


  e      e   2  
   Bi  
3 3 0 ij 2 2 0 ik 2
M
 i 1 i i i
w B D D B D
 j  N
i i i

j 1  D 1  3 e e   3 e e 
2 ik 
      
1ij 2 ij
   2 
1ik
 i 1 w B D  e e
  i  v
3 3 0 ij 2 2 0 ik 2
 i i i D i B i D i B

1  e1ik 3e2 ik  1  3e1ik e2 ik 


    
NB ND
 e0 ik  2  2     e0 ik  2  2 
U v

i 1 Bi Di  Bi  v i 1 Bi Di  Bi  v
k k
Di Di

(k=1,2,,M)

The recursive algorithm developed for reinforced concrete building


design is very much similar to that for tall steel building design. The only
difference is that there are two design variables for each rectangular
concrete member and therefore a more complicated recursive algorithm
is presented. By successively solving for the above recursive relations,
the continuous optimal design solution is then obtained.

Tall Concrete Shear Wall Building

 Shear walls have been the most common types of concrete building
structures in Hong Kong.

 Consider that the shear wall panel members are modelled by rectangular
shell elements

D
L
- 4.31 -

 The same principle of virtual work can be used to formulate the lateral
drift response of a shear wall structures as shown in the following

Constitutive relationship for shell elements

   E  

i.e.
 E E 
 0 0 0 0 
 11  1   1  2
  11 
2

   E E
0 0 0 0   22 
 22  1   2 1  2  
 33   0 0 0 0 0 0   33 
   
 12   0 0 0 G 0 0   12 

 23   0 0 0 0
5
G 0   23 
   6   
 13  5   13 
 0 0 0 0 0 G
 6 
E
in which G  ,  is Poisson’s ratio.
2(1   )

From the stress-strain relationship, the strain states can be rewritten


as
 11   22
 11 
E
 22   11
 22 
E
 33  0

 12  12
G

 23  23
5
G
6

 13  13
5
G
6

Let x3 be the thickness coordinate measured from the midsurface of


the wall. The element internal stresses can be obtained from internal
element forces and moment as
- 4.32 -
F11 12 M 11
 11   x3
t t3
F 12 M 22
 22  22  x3
t t3
 33 0
F12 12 M 12
 12   x3
t t3
F
 13  13
t
F
 23  23
t

where forces Fij and moments M ij are shown per unit of in-plane
length.
- 4.33 -

With explicit stress and strain expressions, any displacement of interest


can be expressed by the principle of virtual work as

  dV
 
 11   22  22   11  12  23  13 
    11   22   12   23   13
5  1
dx dxdx3
E E G 5
 G G
 6 6 
1

 E  11 11   22 22   11 22   22 11 
   dx1dxdx3
 1     6    6    
 G  12 12 5 23 23 5 13 13  

Or,

  dV
LD 1 F f
 12M11m11 F22 f 22 12M 22 m22
     11 11   
0 0 E  t t3 t t3
F f 12M11m22 F f 12M 22 m11 
 11 22     22 11  
t t 3
t t3 
1  F12 f12 12M12 m12  6  V23v23  V13v13  
       dx1dx2
G t t3  5G  t 
 6 6 
F12 f12  V23v23  V13v13 
1L D
F11 f11  F22 f 22   F11 f 22   F22 f11 5 5
     dx1dx2
t 0 0 E G 
 
12  M m  M 22 m22   M11m22   M 22 m11 M12 m12 
L D
 3    11 11   dx1dx2
t 0 0 E G 
- 4.34 -

For a concrete building structure with n shear wall panels, the lateral
drift can be written explicitly in terms of the thickness variables as
n n C C' 
   dV    i  3i 
i 1 i 1  ti ti 

in which
 6 6 
LD F12 f12  V23v23  V13v13 
F11 f11  F22 f 22   F11 f 22   F22 f11 5 5
Ci       dx1dx2
00 E G 
 i
L D M m  M m  M m  M m
 M m 
Ci'  12    11 11 22 22 11 22 22 11
 12 12  dx1dx2
00 E G i

 Using the OC Method to first derive the Optimality Criteria and then apply
the following recursive algorithm to obtain the optimal solution

Recursive relations for ti :

 
1  M  j  eij 3eij   
'

tiv1  tiv 1       1 
 w  j 1 wi  ti2 ti4   
  v 

Recursive relations for  j :

N w    v1 v
Considering  kU   kv    k
 
 ti  ti , one obtains a system of
i 1   ti 

simultaneous linear equations for  j as

D  NW 1  3eij'   3eik' 
 j    ij
e    ik
e   
j 1  i 1 witi 
3
ti2   ti2  
NW 1  3eik' 
=   eik  2   w ( kU   k ) (k=1,2,…, M)
i 1 ti  ti 
- 4.35 -

Research on Cost-Efficiency of Structural Design


of Public Housing Blocks

Commissioned by
the Housing Department of HKSAR Government

Public Housing
Blocks
- 4.36 -

Harmony 1 Block

•Modeling Assumptions as provided:


- rigid floor diaphragm
- material, sectional, geometrical prop.
- severe wind condition
•ETABS analysis
•Wind drift criteria: H/500

Concord Block
- 4.37 -

New Cruciform Block

Summary Results

Block Type Savings in Increase in Reduction in


Const. Cost Usable Floor Dead Weight
(HK$ Million) Area (m 2) (tonne)
Harmony1.1 4.50 855 6489
Concord 1 3.70 300 6745
NCB 1.75 466 3630

Benefits: Disadvantages:
1. Direct savings in material and construction costs 1. Stepped ceiling
2. Increase in usable floor area (0.9-1.3 m 2 per flat) 2. Possible rerouting and refitting service fixtures
3. Better flexibility in the use of floor space 3. Further redesigns are necessary
4. Possible higher sales rate due to larger usable
floor area
5. Increase in floor efficiency ratio of 1.8% for
Concord 1 block
6. Lighter structure may reduce foundation cost
- 4.38 -

Structural Optimization Study for HOS


Development at Shatin Area 52, Shui Chuen O

Shui Chuen O Residential Project Development

• 7 blocks varying height from 32 to


38 storeys
• Total GCFA = 27,894 m 2
• Total number of flat units = 2500
• Couple shear wall construction
- 4.39 -

Structural Optimization Process


Development of Structur al
Form
Initiate Wi nd Engineering
Study
Development of Pr eliminar y
Structural D esign

Preliminar y Optimiz ation

Preliminar y
Recommendations
Results of Wi nd Engineering
Study
Structural Opti mization Using
Wind Engineering Results

Final
FinalRecommendati
Recommendations on for
for
Improvi
Improvi
ng Structural
ng Structural
Efficienc y
Efficienc y

Preliminary Optimization
Review of
Analytical Model
• Only the single block 4 was considered
• Design based on analytically derived wind
Formulation of
loads Optimal Design Problem

• Purpose of this preliminary optimization:


- to review the analysis model
Identification of efficiency
- to identify the most efficient/inefficient of structural elements
elements
- to resize element sizes so as to minimise
structural material cost and to maximise Recommendations for
usable floor area Improvement
- 4.40 -

Optimum Design Problem Formulation


Design Objective:
Minimise structural material cost while maximising usable floor
area

Design constraints:
Lateral drift under all wind conditions  H/500
Min. element size limits = strength-based sizes
No wall is allowed to be thinner than 200 mm
Maximum depth of beams above door opening = 600 mm
Maximum depth of corridor slabs = 300 mm
All lintel beams are sized to have their width to be the same as
the thickness of their adjoining walls

Design variables:
Element sizes at various level zones

Planform of Structure

Typical Floor Plan


- 4.41 -

Identification of Efficiency of Structural Elements

10/F - 12/F

Hotter elements represent


G/F - 3/F more efficient elements to
resist overturning moment
and twisting torsion at the
base levels
- 4.42 -

10/F - 12/F

Hotter elements represent


G/F - 3/F more efficient elements to
resist overturning moment
and twisting torsion at the
base levels

Preliminary Optimization Recommendations

Thicken major shear walls in wings


- 4.43 -

Preliminary Optimization Recommendations

Stiffen the open core walls and deepen core link beams

Preliminary Optimization Recommendations

Add beams to wings to enhance torsional resistance


- 4.44 -

Preliminary Optimisation Recommendations

Reduce most walls to their minimum strength based sizes

Preliminary Optimization Recommendations

Remove inefficient coupling beams and walls in type B1 flat units


- 4.45 -

Detailed Optimisation

• Clustered blocks 1, 2 and 3 and single block 4 were


considered
• Wind loads derived from the wind tunnel test

• Purpose of this detailed optimization:


- to identify the most efficient/inefficient elements under
a more refined set of wind loads
- to finalise the optimum element sizes so as to provide
minimum structure cost and maximum usable floor area

Identification of Efficiency of Structural Elements

Top Levels

Base Levels Behavior is similar to the


stage 1 optimization model
- 4.46 -

Identification of Efficiency of Structural Elements

Base Levels

Clustered blocks 1, 2 & 3 are linked


together to resi st wind such that the
main web walls of the end wings
become the most efficient elements

Identification of Efficiency of Structural Elements

Top Levels

All walls could be reduced to the


minimum strength ba sed sizes
- 4.47 -

Two Optimisation Schemes

Zone 2
(20/F - Roof )

Scheme 1 Scheme 2
Uniform wall Two wall
thickness thickness
variations

Zone 1
(Base - 20F)

Scheme 2 Optimisation Model


Work Contribution
per Floor
UROOF2

ROOF

32

30
Zone 2
(20/F - Roof ) 28

26

24

22

20

18

16

14

12

Zone 1 10
(Base - 20F) 8

P4

0% 5% 10% 15% 20%


- 4.48 -

Optimum Wall Sizes for Block 4


Scheme 1 - Uniform Wall Thickness Along Height

Optimum Wall Sizes for Block 4


Scheme 2 - Two Wall Thickness Variations Along Height

Zone 1: GF - 20F
Wall sizes same as option 1 results

Zone 2: 21F - Roof


All wing walls are reduced to min.
200 mm thick
- 4.49 -

Optimum Wall Sizes for Blocks 1, 2 & 3


Scheme 1 - Uniform Wall Thickness Along Height

All walls are reduced to minimum


strength based sizes

Optimum Wall Sizes for Blocks 1, 2 and 3


Scheme 2 - Two Wall Thickness Variations Along Height

Zone 1: GF - 20F
Wall sizes same as option 1 results
Zone 2: 21F - Roof
All wing walls are reduced to min.
200 mm thick
- 4.50 -

Optimization Results for Block 4


Savings in Concrete Volume

0.40
Vol. of Concrete per CFA (m3/m2)

0.35
0.349
0.30 0.307
(-12%) 0.283
0.25 0.267
(-19%)
(-24%)
0.20

0.15

0.10

0.05

0.00
Prelim. Stage 1 Opt. Stage 2 Opt. Stage 2 Opt.
Design - Scheme 1 - Scheme 2

Optimization Results for Block 4


Benefit of More Usable Floor Area

14%
Area Occupied by Walls

12%
11.4%
10% 10.7%
9.9%
(-6%)
per CFA

9.3%
8% (-13%)
(-18%)
6%

4%

2%

0%
Prelim. Stage 1 Opt. Stage 2 Opt. Stage 2 Opt.
Design Scheme 1 Scheme 2
- 4.51 -

Optimization Results for Blocks 1, 2 &3

Savings in Concrete Volume

0.40

0.35
Vol. of Concrete per CFA

0.30 0.310
0.280
0.25 0.272
(m3/m2)

(-10%)
(-12%)
0.20
0.15

0.10

0.05

0.00
Stage 1 - Stage 2 Opt. - Stage 2 Opt. -
Initial Design Scheme 1 Scheme 2

Optimization Results for Blocks 1, 2 & 3

Benefit of More Usable Floor Area

14%
Area Occupied by Walls per CFA

12%

10% 10.2%
9.2% 8.9%
8%
(-10%) (-13%)
6%

4%

2%

0%
Stage 1 - Stage 2 Opt. Stage 2 Opt.
Initial Design Scheme 1 Scheme 2
- 4.52 -

Summary of Stage 2 Optimization Results for Blocks 1, 2, 3 & 4

Scheme 1: Uniform Wall Thickness


• Total savings in volume of concrete = 3220 m 3 (6% total vol. of conc)
• Total increase in usable floor area = 1090 m 2 (1% CFA )

Scheme 2: Two Wall Thickness Variations


• Total savings in volume of concrete = 4390 m 3 (8% total vol. of conc)
• Total increase in usable floor area = 1510 m 2 (1.3% CFA )

Summary of Stage 2 Optimization Results for all blocks 1 - 7

Scheme 1: Uniform Wall Thickness


• Estimated total savings in vol. of conc rete = 5630 m 3 (6% total vol. of conc)
• Estimated total increase in usable floor area = 1900 m 2 (1% CFA)

Scheme 2: Two Wall Thickness Variations


• Estimated total savings in vol. of conc rete = 7650 m 3 (8% total vol. of conc)
• Estimated total increase in usable floor area = 2640 m 2 (1.3% CFA)
- 4.53 -

Kowloon Station Development


The Harbourside

Sponsored by Hung Lung Properties Ltd. in


conjunction with Ove Arup & Partners Hong Kong Ltd.

The Harbourside Development

Harbourside
Development

Courtes y of Ove Arup & Partners


- 4.54 -

Package 7

Packages 5 & 6

Package 2
The
H ar
bou
rsid
e

Package 1
Package 3

Site Plan of Kowloon Station Development

 Three 67-storey residential towers above a 6-level


podium
 Building height = 242.5 m
 Plan dimensions = 133 m x 25.5 m
 Height-w idth aspect ratio = 9.5:1
 Totally 1135 flat units and 133,476 m2 GFA
 Coupled shear wall reinforced concrete
construction
 Minimize structure cost with maximum usable
floor area
 Design constraints:
Overall building deflection  H/500
Interstorey drif t  h/300 (diagonal w ind)
 h/350 (perpendicular w ind)
Min. element sizes = strength-based sizes
Max. element sizes specif ied by the client
 Design variables:

Harbourside Development element sizes at various level zones


- 4.55 -

Optimization Challenges
• Large-scale structure
5864 frame elements
25528 wall panel elements
• Strongly coupled lateral-torsional
motion

• 24 wind load combinations

• Multiple control of lateral drifts

• Variable unit costs, grade of concrete


and stiffness enhancement by steel
reinforcement

Structural Modeling
- 4.56 -

High-rise Level Mid-rise Level

Low-ris e Level Podium Level

Lateral Wind Drift Performance Before Optimization


Critical Wind Load Combination: -0.75Fx+0.85Fy+0.65 Mz
- 4.57 -

Lateral Wind Drift Performance Before Optimization

Design Optimization History

106000
104000 11% savings in concrete = 12340 m3
Volume (m 3)

102000
100000
98000
96000
94000
92000
90000
0 1 2 3 4 5 6 7
Design Cycle
- 4.58 -

Relative Work Contribution Under Most Critical Wind Load Case

Highrise levels

Lowrise levels

Podium levels

(Note: Hotter color denotes elements of higher eff iciency )

Key Recommendations:

On the podium structure


• Recommend to reduce the transfer plate thickness from 5.5 m to
its minimum strength-based thickness of 5 m
• Recommend to limit the width of podium columns to 4 m but to
increase the depth from 2.4 m to 3 m
• Recommend that the concrete grade of podium columns be
modified to C60 with stiffness enhancement by 5% steel
reinforcement
- 4.59 -

On the tower structure

3/F
Original

• Recommend thickening most end


walls especially those at the
outermost locations

1st Round
Optimization

Original 33/F
• Recommend widening coupling
beams around the main core
• Recommend reducing the
thickness of most walls

 Recommend thickening back


walls at re-entrants

1st Round
Optimization
- 4.60 -

Original 63/F
• Recommend reducing the
thickness of all walls
• Recommend widening coupling
beams around the main core

1st Round
Optimization

Summary Results
• Total savings in concrete volume = 10416 m3 (11% of initial concrete volume)
• Total savings in usable floor area = 2610 m2 (26 flat units)
• Average increase in usable floor area per flat = 2.5 m2
• Equivalent increase in floor efficiency = 2.2%

Assuming a composite unit cost of HK$2000/m 3 for reinforced concrete and a unit
price of HK$60,000/m2 for floor area
• Total estimated cost savings in concrete = HK$ 20.8 M
• Total equivalent cost savings in floor areas = HK$ 156.6 M
• Total cost savings = HK$ 177.4 M
• Additional financial savings in the foundation due to the reduction of concrete usage
- 4.61 -

Results of 7 Highrise Concrete Buildings Optimized


Benefits of Concrete Material Savings

Before Optimisation After Optimisation

0.9 6
0.8
Vol. of concrete per GFA (m3/m2)

0.7
5 7

0.6 3 4

0.5
1
0.4 2

0.3

0.2

0.1

0
100 200 300
Height (m)

Results of 7 Highrise Concrete Buildings Optimized


Benefits of Increase in Usable Floor Space

Before Optimisation After Optimisation

6 7
16%
Walls/Columns per GFA (m2/m2)

4
14% 5
1 2 3
Area of Occupied by

12%
10%
8%
6%
4%

2%
0%
100 150 200 250 300
Height (m)

You might also like