Académique Documents
Professionnel Documents
Culture Documents
Course Details
Visiting
Lecturers
Dr. Ken Doust
Mr Ian McIntyre
Demonstrators Melissa Duell Email: m.duell@unsw.edu.au
Milad
Ghasrikhouzani
Email: m.ghasrikhouzani@unsw.edu.au
Yanni Huang Email: yanni.huang@unsw.edu.au
Divya J Nair Email:
d.jayakumarnair@student.unsw.edu.au
Neeraj Saxena Email: n.saxena@student.unsw.edu.au
Kelly Tang Email: kellyjtang@gmail.com
Kasun Wijayaratna Email: z3188830@unsw.edu.au
Koray Yurt Email: k.yurt@unsw.edu.au
http://www.civeng.unsw.edu.au/info-about/our-school/policies-
procedures-guidelines/academic-advice
Reading
Todays Lecture (Larson and Odoni, 1981)
Sections 1.0, 1.1, 1.2 and 1.3
Todays Lecture (Penn and Parker, 2011)
Chapter 1
Upcoming Material (Larson and Odoni, 1981):
Section 6 (Intro), 6.1, 6.2, 6.2.1
Minimum-Spanning-Tree: Section 6.3, 6.3.1
Facility Location: Section 6.5, 6.5.1 through 6.5.8
Suggested as core background for the course:
AMPL Textbook:
Introduction, Chapter 1, and Chapter 2
Course Overview
Domain knowledge about infrastructure
Facts & figures
Standards
Organizational processes
Technical knowledge about systems modelling and optimization
Mathematical techniques
Algorithms
Software usage
Intentional Student Challenges:
We will move back and forth between these types of knowledge.
We will rely frequently on word problems. You must develop the
mathematical model.
Undergraduate Research Assistance
In the past Ive often hired a couple Undergraduate RAs.
Application would be late in the course with any decision just after
the course ends.
Activities would include involvement in research, governmental
and/or industry projects.
Will provide more details mid-semester.
This slide is simply to give some notice to those who may be
interested and as a reminder to myself
Some Infrastructure Disciplines Highly Relevant
for Engineering
Energy
Water
Transport
Telecommunications
Basically, the foundational necessities for a properly
functioning society
The Role of Engineers
We are not technicians
We are problem solvers
We combine our technical and domain knowledge, often in novel ways
We must be rigorous
We must adapt as needed
Note: anything completely repetitive will (eventually) be replaced by a
computer
The Role of Engineers
From Penn and Parker (2011)
Increased Demands on Urban Infrastructure
Penn and Parker (2011)
Infrastructure Management is not only Design and Construction
Penn and Parker (2011)
Common Infrastructure Activities by Budget Range
Penn and Parker (2011)
Closer to home
Suggested side reading. Chapter 1 and Chapter 10 (Funding) of the
NSW Long Term Transport Master Plan:
http://www.transport.nsw.gov.au/sites/default/files/b2b/publications/n
sw-transport-masterplan-final.pdf
Billions in transport infrastructure
2010-2011, $11.7
2011-2012, $13.1
2012-2013, $13.2
Transport investment represents around 40 percent of total State
infrastructure investment
Funding Sources (NSW Transport Master Plan)
State Organizational
Structure for Transport
Infrastructure
(NSW Transport Master Plan)
Urban Service Costs Larson and Odoni (1981)
Systems Modelling - Larson and Odoni (1981)
This is just one approach to some
Systems Modelling problems
Systems Modelling
What is a systems model?
Is a mathematical model ever perfect?
What is a good mathematical model?
Applied Systems Modelling
In this course, we will apply such systems thinking to infrastructure
applications.
To address infrastructure applications, we will categorize them into
stages.
One View of the Stages of Infrastructure Systems Optimization
Planning
What to build
Where to build
Finance
Who pays
Construction
How to build
When to build
Operations
What to do day-to-day/month-to-month
Maintenance
What to fix
When to replace
Short Motivational Example of Transport
Infrastructure Planning (Braess Paradox)
Should we build a bridge at a particular location?
Many considerations
Cost
Disruption during construction
Environmental Impact
Impact on accessibility, travel time, safety, etc etc.
This is a grossly simplified example but grounded in a real issue.
Let us only consider the expected benefit side, specifically travel
time savings.
Simplified Static Equilibrium Model
Braesss Paradox (1950s simplified example)
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
P
1
= A-B-D
P
2
= A-C-D
2 Paths
Simplified Static Equilibrium Model
Braesss Paradox (1950s simplified example)
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
P
1
= P
2
= 25
Equilibrium flows
P
1
= A-B-D
P
2
= A-C-D
2 Paths
c
1
+c
2
= c
3
+c
4
= 9.5
Total cost = 475
Braesss Paradox Example
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
1
5
c
P
1
= A-B-D
P
2
= A-C-D
P
3
= A-B-C-D
3 Paths
Braesss Paradox Example
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
1
5
c
P
1
= A-B-D
P
2
= A-C-D
P
3
= A-B-C-D
3 Paths
5
1
c
5
4
c
Braesss Paradox Example
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
1
5
c
P
1
= A-B-D
P
2
= A-C-D
P
3
= A-B-C-D
3 Paths
P
3
= 50, P
1
= 0, P
2
= 0
Equilibrium flows
C
1
+ C
5
+ C
4
= 11
Total cost = 550
Braesss Paradox Example
A
B
C
D
10
1
1
x
c
10
4
4
x
c 7
3
c
7
2
c
X=50
1
5
c
P
1
= A-B-D
P
2
= A-C-D
P
3
= A-B-C-D
3 Paths
P
3
= 50, P
1
= 0, P
2
= 0
Equilibrium flows
C
1
+ C
5
+ C
4
= 11
Total cost = 550
Bottom Line: We would likely be spending millions
of Dollars to build a bridge that made everyones
travel time worse
One View of the Stages of Infrastructure Systems Optimization
Planning
What to build
Where to build
Finance
Who pays
Construction
How to build
When to build
Operations
What to do day-to-day/month-to-month
Maintenance
What to fix
When to replace
We will return to infrastructure knowledge
in the context of these stages.
But, before that, we will begin to address the
Mathematical tools we will use in addressing
these stages
Mathematical Modelling/
Linear Programming
Variables
The decisions we have control over.
Objective function
A function representing profit, cost, consumption or some other value to
be maximized or minimized.
Constraints
A set of restrictions placed on the variables.
Non-negativity constraints
A special set of constraints stating that a variable cannot be negative.
Mathematical Modelling/
Linear Programming
Variables
The decisions we have control over. Example: x, y, z or x
1
, x
2
, x
3
Objective function
A function representing profit, cost, consumption or some other value to
be maximized or minimized. Example: f(x) or f(x
1
, x
2
, x
3
) or f(x,y,z)
Constraints
A set of restrictions placed on the variables.
Examples x 10 or x
2
x
3
= 5 or y + z 13
Non-negativity constraints
A special set of constraints stating that a variable cannot be negative.
Example, x
1
0
For a math model, if the objective function is linear, all constraints
are linear and all variables are continuous, then the model is a
Linear Program (LP)
LPs are generally good because we have general solution
methods to efficiently solve them.
If the variables must take on integer values then the model is an
Integer Program (IP) when all variables are integers or Mixed
Integer Program (MIP) when some variables are integers and
comes are continuous.
Mathematical Modelling/
Linear Programming
Linear Programs (from the AMPL textbook)
An (extremely simplified) steel company must decide how to allocate next weeks time on a rolling
mill. The mill takes unfinished slabs of steel as input and can produce either of two semi-finished
products, which we will call bands and coils. The mills two products come off the rolling line at
different rates:
Tons per hour: Bands 200 Coils 140
They have different profitabilities: Bands $25 Coils $30
The following weekly production amounts are the most that can be justified in light of the currently
booked orders: Bands 6,000 Coils 4,000
If we have 40 hours of production time available this week, how many tons of bands and how
many tons of coils should be produced to bring in the greatest total profit?
Formulating the LP
Multiple ways to begin.
We will adopt this process:
1. Identify the decision variables in words
2. Explicitly restate the decision variable in mathematical notation
3. Identify the constraints in words
4. Explicitly restate the constraints in mathematical notation
5. Identify the objective in words
6. Explicitly restate the objective in mathematical notation
Some problems may lend themselves to different orderings of these
steps.
Linear Programs (from the AMPL textbook)
An (extremely simplified) steel company must decide how to allocate next
weeks time on a rolling mill. The mill takes unfinished slabs of steel as input
and can produce either of two semi-finished products, which we will call
bands and coils. The mills two products come off the rolling line at different
rates:
Tons per hour: Bands 200 Coils 140
They have different profitabilities: Bands $25 Coils $30
The following weekly production amounts are the most that can be justified
in light of the currently booked orders: Bands 6,000 Coils 4,000
If we have 40 hours of production time available this week, how many tons
of bands and how many tons of coils should be produced to bring in the
greatest total profit?
LP Example
Decision Variables?
LP Example
Decision Variables
Tons of Bands to produce
Tons of Coils to produce
LP Example
Decision Variables
Tons of Bands to produce X
B
Tons of Coils to produce X
C
LP Example
Constraints?
LP Example
Constraints
Cannot produce for longer than 40 hours
Cannot make more than 6000 tons of Bands
Cannot make more than 4000 tons of Coils
LP Example
Constraints
Cannot produce for longer than 40 hours
Cannot make more than 6000 tons of Bands
Cannot make more than 4000 tons of Coils
Anything else?
LP Example
Constraints
Cannot produce for longer than 40 hours
Cannot make more than 6000 tons of Bands
Cannot make more than 4000 tons of Coils
Cannot make negative tons of Bands
Cannot make negative tons of Coils
LP Example
Constraints
Cannot produce for longer than 40 hours
Cannot make more than 6000 tons of Bands X
B
6000
Cannot make more than 4000 tons of Coils X
C
4000
Cannot make negative tons of Bands X
B
0
Cannot make negative tons of Coils X
C
0
LP Example First Constraints?
Cannot produce for longer than 40 hours
Tons per hour: Bands 200 Coils 140
Each ton of Bands takes 1/200
th
of an hour
Each ton of Coils takes 1/140
th
of an hour
So, the constraints is
(1/200) X
B
+ (1/140) X
C
40
LP Example
Constraints
Cannot produce for longer than 40 hours
(1/200) X
B
+ (1/140) X
C
40
Cannot make more than 6000 tons of Bands X
B
6000
Cannot make more than 4000 tons of Coils X
C
4000
Cannot make negative tons of Bands X
B
0
Cannot make negative tons of Coils X
C
0
LP Example
Objective?
LP Example
Objective
Maximize Profit
LP Example
Objective
Maximize Profit 25 X
B
+ 30 X
C
Overall LP
Maximize 25 X
B
+ 30 X
C
Subject to
(1/200) X
B
+ (1/140) X
C
40
0 X
B
6000
0 X
C
4000
Overall LP
Maximize 25 X
B
+ 30 X
C
Subject to
(1/200) X
B
+ (1/140) X
C
40
0 X
B
6000
0 X
C
4000
This would be the answer to the question posed on the earlier slide.
However, this alone would not necessarily warrant full credit. The development of the
answer is also important. Specify the variables, constraints, and objective in your own
words first. Then the mathematics.
Coded for AMPL (saved as prod0.mod)
var XB;
var XC;
maximize Profit: 25 * XB + 30 * XC;
subject to Time: (1/200) * XB + (1/140) * XC <= 40;
subject to B_limit: 0 <= XB <= 6000;
subject to C_limit: 0 <= XC <= 4000;
Coded for AMPL (saved as prod0.mod)
var XB;
var XC;
maximize Profit: 25 * XB + 30 * XC;
subject to Time: (1/200) * XB + (1/140) * XC <= 40;
subject to B_limit: 0 <= XB <= 6000;
subject to C_limit: 0 <= XC <= 4000;
To find the solution:
ampl: model prod0.mod;
ampl: solve;
ampl: display XB, XC;
Reminders
Please read course profile closely. You are responsible for knowing it.
Other reading:
Todays Lecture (Larson and Odoni, 1981)
Sections 1.0, 1.1, 1.2 and 1.3
Todays Lecture (Penn and Parker, 2011)
Chapter 1
Upcoming Material (Larson and Odoni, 1981):
Section 6 (Intro), 6.1, 6.2, 6.2.1
Minimum-Spanning-Tree: Section 6.3, 6.3.1
Facility Location: Section 6.5, 6.5.1 through 6.5.8
Suggested as core background for the course:
AMPL Textbook:
Introduction, Chapter 1, and Chapter 2