2. Course Objectives
โข Learn basic optimization methods and how they are applied in
engineering design
โข Use MATLAB to solve optimum engineering design problems
โ Linear programming problems
โ Nonlinear programming problems
โ Mixed integer programing problems
3. Course Materials
โข Arora, Introduction to Optimum Design, 3e, Elsevier,
(https://www.researchgate.net/publication/273120102_Introductio
n_to_Optimum_design)
โข Parkinson, Optimization Methods for Engineering Design, Brigham
Young University
(http://apmonitor.com/me575/index.php/Main/BookChapters)
โข Iqbal, Fundamental Engineering Optimization Methods, BookBoon
(https://bookboon.com/en/fundamental-engineering-optimization-
methods-ebook)
4. Engineering Design Examples
โข Soda can design
โข Insulated spherical tank
โข Two-bar truss
โข Three-bar overhead truss
โข Cantilever column
โข Stepped cantilever beam
โข Coil spring
โข Heat pump
5. Design Example: Soda Can Design
Problem: A soda can of 125 ๐๐ capacity is to be designed to minimize
material costs (proportional to surface area. For esthetic reasons,
the height to diameter ratio should be at least 2.
Design variables: height (โ), diameter (๐)
Objective: ๐๐๐โ,๐ ๐ =
1
2
๐๐2
+ ๐๐โ
Subject to: ๐1:
1
4
๐๐2โ โ 125 = 0 (volume)
๐2: 2๐ โ โ โค 0
Trial design:
Let ๐ = 4๐๐, โ = 8๐๐
Then ๐ = 125.66 ๐๐2
๐1 ๐ฅ = โ24.47; ๐2 ๐ฅ = 0
6. Design Example: Insulated Spherical Tank
Problem: choose the insulation thickness (๐ก) to minimize the life-cycle
costs of a spherical tank of radius ๐ .
Life cycle costs: ๐2๐ด๐ก + ๐3๐บ + ๐4๐บ โ ๐๐ค๐
Annual heat gain: ๐บ = 365 ร 24 ร ฮ๐ ร
๐ด
๐ร๐ก
Surface area: ๐ด = 4๐๐ 2 [๐2
]
Thermal resistivity: ๐ ๐ โ ๐ ๐๐ โ ยฐ๐ถ/๐ฝ
Equipment insulation cost: ๐2 $/๐3
Equipment refrigeration cost: ๐3 $/๐โ
Annual operating cost: ๐4 $/๐โ
Present worth factor: ๐๐ค๐ =
๐ด
๐
1 โ
1
1+๐ ๐
Note, there are no constraints in this problem
7. Design Example: Rectangular Beam
Problem: design a rectangular beam of minimum cross-section area to
withstand specified bending moment (๐) and shear load (๐).
Design variables: width (๐), depth (๐)
Objective: ๐ ๐, ๐ = ๐๐
Constraints:
Bending stress: ๐ =
6๐
๐๐2 โค ๐๐
Shear stress: ๐ =
3๐
2๐๐
โค ๐๐
Aspect ratio:
๐
๐
โค 2
Problem specific data
๐ = 140 ๐๐; ๐ = 24 ๐๐;
๐๐ = 165 ๐๐๐; ๐๐ = 50๐๐๐
8. Design Example: Hollow Cantilever Beam
Problem: design a hollow cantilever beam of length ๐ฟ of square cross-
section to support a load ๐ with minimum mass (Ref: Arora, p. 18)
Design variables: width ๐ค , thickness ๐ก
Objective: minimum ๐๐๐ ๐ = ๐๐๐ด [๐พ๐]
Cross-section area: ๐ด = 4๐ก ๐ค โ ๐ก [๐๐2]
Moment of inertia: ๐ผ =
1
12
๐ค4 โ ๐ค โ 2๐ก 4 [๐๐4]
Moment about neutral axis: ๐ =
1
8
๐ค3 โ ๐ค โ 2๐ก 3 [๐๐3]
Bending stress: ๐ =
๐๐๐ค
2๐ผ
[๐ โ ๐๐โ2]
Shear stress: ๐ =
๐๐
2๐ผ๐ก
[๐ โ ๐๐โ2]
End deflection: ๐ =
๐๐ฟ3
3๐ธ๐ผ
[๐๐]
12. Design Example: Coil Spring
Problem: design a minimum mass spring to carry a given axial load ๐ without
material failure while satisfying minimum deflection and minimum surge
wave frequency requirements
Design variables: mean coil diameter ๐ท , wire diameter ๐ , number of
active coils (๐)
Design equations:
Spring mass: ๐ =
1
4
๐ + ๐ ๐2
๐ท๐2
๐
Load deflection: ๐ = ๐พ๐ฟ, where ๐พ =
๐4๐บ
8๐ท3๐
Shear stress: ๐ =
8๐๐๐ท
๐๐3
Stress concentration factor: ๐ =
4๐ทโ๐
4(๐ทโ๐)
+ 0.615
๐
๐ท
Frequency of surge waves: ๐ =
๐
2๐๐๐ท2
๐บ
2๐
14. Design Example: Stepped Cantilever Beam
A fixed load ๐ = 50๐๐ is supported at the end of a stepped cantilever
beam of ๐ฟ = 500๐๐. The sections are of equal length (๐ = 100๐๐).
The design objective is to minimize the mass of the beam by
reducing its total volume.
Design variables: ๐๐, โ๐, ๐ = 1, โฆ , 5
Objective: min ๐ = ๐ ๐1โ1 + ๐2โ2 + ๐3โ3 + ๐4โ4 + ๐5โ5
16. Design Example: Symmetric Two-Bar Truss
Problem: a symmetrical two-bar truss is to be designed with hollow
tubes to support a fixed load ๐. The truss has height ๐ป, and span ๐ต.
Objective: minimum weight structure that will withstand ๐, and not
yield, buckle, or deflect excessively.
Design variables: diameter ๐ , thickness ๐ก
Alternately, choose: diameter ๐ , height ๐ป
๐ = (๐ต/2)2+๐ป2, ๐ด = ๐๐๐ก
Total weight: ๐ = 2๐๐๐ด,
Axial stress: ๐ =
๐๐
2๐๐๐ก๐ป
Buckling stress: ๐๐ =
๐2 ๐2+๐ก2 ๐ธ
8๐2
Deflection: ๐ =
๐๐3
2๐๐๐ก๐ป2๐ธ
17. Design Example: Symmetric Two-Bar Truss
The optimum design problem is defined as:
Objective: min
d,t
๐ = 2๐๐๐ก๐ (๐ต/2)2+๐ป2
Subject to: ๐๐ โค ๐ โค ๐๐; ๐ โค ๐๐๐๐ฅ;
Simplify objective and scale the constraints:
Objective: min
d,t
๐ = ๐ ร ๐ก
Subject to:
๐๐
๐
โ 1 โค 0,
๐
๐๐
โ 1 โค 0 ,
๐
๐๐๐๐ฅ
โ 1 โค 0,
For a particular problem, let ๐ = 66,000 ๐๐; ๐ป = 30 ๐๐; ๐ต = 60 ๐๐;
๐ = 0.3
๐๐
๐๐3
; ๐ธ = 30 ร 106
๐๐
๐๐2
; ๐๐ = 1 ร 105
๐๐ ๐; ๐๐๐๐ฅ = 0.25 ๐๐
Then ๐ = 79.97 ๐ ร ๐ก, ๐ = 14,855 ๐ ร ๐ก, ๐๐ = 20,562 ๐2
+ ๐ก2
, ๐ =
0.0297
๐ร๐ก
;
18. Design Example: Three-Bar Overhead Truss
Problem: A minimum-weight symmetric three-bar overhead truss is to
be designed to support a load applied at an angle (Ref: Arora, p. 46).
The truss has height ๐ and span 2๐; member 1 and 3 have length 2๐
and area ๐ด1; member 2 has length ๐ and area ๐ด2.
Load ๐ is applied at the joint at an angle ๐; the horizontal and vertical
components of the applied load are: ๐๐ข = ๐ cos ๐ , ๐๐ฃ = ๐ sin ๐.
Design variables:
cross-sectional areas, ๐ด1 and ๐ด2
Design objective:
minimize ๐ = ๐๐๐(2 2๐ด1 + ๐ด2)
Design constraints:
Stress, deflection, buckling, resonance
19. Design Example: Three-Bar Overhead Truss
The stresses in the three structural members are given as:
๐1 =
1
2
๐๐ข
๐ด1
+
๐๐ฃ
(๐ด1+ 2๐ด2)
; ๐2 =
2๐๐ฃ
(๐ด1+ 2๐ด2)
; ๐3 =
1
2
โ
๐๐ข
๐ด1
+
๐๐ฃ
(๐ด1+ 2๐ด2)
The horizontal and vertical deflections of the load point are:
๐ข =
2๐๐๐ข
๐ด1๐ธ
, ๐ฃ =
2๐๐๐ฃ
(๐ด1+ 2๐ด2)๐ธ
The buckling load for the members in compression is given as:
๐2๐ธ๐ผ
๐๐
2 ,
where the moment of inertia is: ๐ผ๐ = ๐ฝ๐ด๐
2
, where ๐ฝ = constant.
The lowest resonance frequency of the structure is given as:
๐ =
3๐ธ๐ด1
๐๐2(4๐ด1+ 2๐ด2)
, where ๐ is the mass density.
22. Design Example: Heat Pump
A heat pump is to be designed to extract residual heat from outside air
and supply it indoor. Electric resistance heating is used as backup.
Design variables: ๐ก๐, ๐ก๐, ๐ก๐๐๐, ๐ก๐๐๐
23. Design Example: Heat Pump
The design problem is to choose the size of the compressor, condenser
and evaporator to minimize the life-cycle costs over 10 years.
The following assumptions are made:
โ The average outside air temperature is 0ยฐ๐ถ
โ Evaporator inlet air temperature is 24ยฐ๐ถ
โ The approach temperature in the exchanger should be at least 4ยฐ๐ถ.
โ The airflow rate through the coils is ๐ = 5 ๐๐/๐ ๐๐
โ The pump operates for 4000 hours every year
โ Compressor cost is $220/๐๐ of motor
โ Cost of coils is $100/๐2
of the air-side area
โ Power cost is $0.10/๐พ๐โ๐
24. Design Example: Heat Pump
The variables are defined as:
Condenser inlet air temperature, ๐ก๐๐๐ = 0ยฐ๐ถ
Condenser outlet air temperature, ๐ก๐๐๐
Evaporator inlet air temperature, ๐ก๐๐๐ = 24ยฐ๐ถ
Evaporator outlet air temperature, ๐ก๐๐๐
Temperature of approach for condenser, ๐ก๐๐๐๐ = ๐ก๐ โ ๐ก๐๐๐
Temperature of approach for evaporator, ๐ก๐๐๐๐ = ๐ก๐๐๐ โ ๐ก๐
Heat transferred in condenser, ๐๐ = ๐๐ถ๐ ๐ก๐๐๐ โ ๐ก๐๐๐ = ๐๐ด๐ฮ๐๐
Heat transferred in evaporator, ๐๐ = ๐๐ถ๐ ๐ก๐๐๐ โ ๐ก๐๐๐ = ๐๐ด๐ฮ๐๐
Log mean temperature difference across condenser,
ฮ๐๐ =
๐ก๐โ๐ก๐๐๐ โ ๐ก๐โ๐ก๐๐๐
ln ๐ก๐โ๐ก๐๐๐ / ๐ก๐โ๐ก๐๐๐
Log mean temperature difference across evaporator,
ฮ๐๐ =
๐ก๐๐๐โ๐ก๐ โ ๐ก๐๐๐โ๐ก๐
ln ๐ก๐๐๐โ๐ก๐ / ๐ก๐๐๐โ๐ก๐
25. Design Example: Heat Pump
Variable definitions:
Heat transfer coefficient, ๐ = .025
๐๐
๐2ยฐ๐ถ
Heat supplied by the electric heater, ๐๐ = ๐๐ถ๐ 35 โ ๐ก๐๐๐
Electrical energy input to the compressor, ๐ = ๐๐ โ ๐๐
Coefficient of performance for compressor,
๐ถ๐๐ =
๐๐
๐
= 7.24 + 0.352๐ก๐ โ 0.096๐ก๐ โ 0.0055๐ก๐๐ก๐
Capital cost of the heat pump, ๐ถโ๐ = 100 ๐ด๐ + ๐ด๐ + 220๐
Operating cost of the heat pump, ๐ถ๐โ๐ = 0.06 ร 4000 ร ๐๐๐น ร ๐
Operating cost of the heater, ๐ถ๐๐โ = 0.06 ร 4000 ร ๐๐๐น ร ๐๐
Present worth factor, ๐๐๐น(๐ด, ๐) =
๐ด
๐
1 โ
1
1+๐ ๐
Total cost, ๐ถ๐ก๐๐ก๐๐ = ๐ถโ๐ + ๐ถ๐โ๐ + ๐ถ๐๐โ
26. Design Example: Heat Pump
Using three independent variables, an unconstrained optimization
problem is defined as:
Objective: min ๐ถ๐ก๐๐ก๐๐(๐ก๐, ๐ก๐, ๐ก๐๐๐)
Alternatively, the constrained optimization problem is defined as:
Objective: min ๐ถ๐ก๐๐ก๐๐(๐ก๐, ๐ก๐, ๐ก๐๐๐, ๐ก๐๐๐)
Subject to: ๐๐ โ ๐๐ โ ๐ = 0
Variable bounds:
35ยฐ๐ถ โค ๐ก๐ โค 70ยฐ๐ถ, โ20ยฐ๐ถ โค ๐ก๐ โค 20ยฐ๐ถ,
0ยฐ๐ถ = ๐ก๐๐๐ โค ๐ก๐๐๐ โค 35ยฐ๐ถ, 0ยฐ๐ถ โค ๐ก๐๐๐ โค ๐ก๐๐๐ = 24ยฐ๐ถ
27. Graphical Optimization Method
โข Before delving into the formal optimization methods, letโs examine
the graphical optimization method, which is applicable for problems
formulated with one or two variables
โข The graphical method involves the following steps:
โ Establish a (2D) grid for variable ranges considered
โ Plot the constraint boundaries and establish the feasible region
โ Using the grid plot contours (level curves) of the cost function
โ Identify the minimum by inspection
28. Graphical Method for Problems in Two Variables
โข Assume that the optimization problem is given as:
Objective: min
๐ฅ1,๐ฅ2
๐(๐ฅ1, ๐ฅ2)
Subject to: ๐๐ โค 0, โ๐ = 0
Variable bounds: ๐ฅ๐
๐ฟ
โค ๐ฅ๐ โค ๐ฅ๐
๐
, ๐ = 1,2
โข We use MATLAB to
โ Define a grid of pairs of values over the ranges of variables
โ Plot the constraint boundaries and locate the feasible region
โ Plot the contours of the objective function to graphically locate a
minimum