MS&E 250A

Engineering Risk Analysis

Instructor:

Professor Elisabeth Pat-Cornell

Huang Engineering Center, Room 336
Phone: (650) 725-1624
Office Hours: Tuesdays, 2pm-3pm
E-mail address: mep@stanford.edu

Location:

Thornton 102 or SCPD

Class times:

Mondays and Wednesdays 1:30-2:50 p.m.

Homework:

Preliminary homework due Monday, January 11. The preliminary homework does not
count for the final grade. There will be four graded homework assignments.

Exams:
Final:

Midterm:
Wednesday, February 1, 1:30-2:50 p.m. (in class)
Monday, March 14, 12:15-3:15 p.m. Location TBD

Course Grade: 45%

30%
25%
+5% bonus

- Final Exam
- Midterm
- Homework
- Attendance and participation (SCPD students may participate via the
discussion forum on Coursework)
Books:
No required textbook. (There is no single text that covers the course material). Class
notes and readings are sufficient. Exercises will be presented in the problem sessions.
Class Notes: Required. Class Notes available at the Stanford Bookstore. One copy is on reserve at the
Engineering Library. Additional readings will be posted on Coursework.
Course
Gregory Heon (gregheon@stanford.edu)
Assistants:
Office Hours: SCPD Google hangout. See Piazza for time.
Matthew Smith (msmith7@stanford.edu)
Office Hours: Fridays, 3-4pm, Huang 219
Philip Keller (pjkeller@stanford.edu)
Office Hours: Mondays, 3-4pm, Huang B011
Marshall Kuypers (mkuypers@stanford.edu)
Office Hours: Tuesdays, 1:30-2:30pm, Huang B011
Problem
Sessions:

Thursdays 3:30 4:20 pm, at Skilling Auditorium. Available through SCPD.

Website:

Enroll via http://coursework.stanford.edu (in addition to registering for the class on

Axess). The course website will include all handouts (downloadable), announcements,
and clarifications to assignments.
Course discussion forum for students: https://piazza.com/stanford/winter2016/mse250a

Other:

Please arrive to class on time. Do not use laptops during class, and please turn off your
cell phones. Attendance is mandatory, and participation will count toward your final
grade through bonus points that can change your letter grade. For this reason, access to
SCPD videos will not be granted to students in residence.

WEEKLY SCHEDULE
Monday

Wednesday

1:30-2:50 p.m. Thornton 102

1:30-2:50 p.m. Thornton 102

Thursday
2:30-3:20pm Review Session
Skilling Auditorium

CALENDAR (GUIDELINES ONLY- ORDER OF TOPICS MAY CHANGE)

Monday

Wednesday

Part 1: BASIC CONCEPTS

Jan

1: Managing and reducing risks

4 2: Risk analysis procedures and examples
Preliminary Homework Assigned

3: Risk characterization and measures

4: Treatment of uncertainty in risk analysis
6
5: Bayesian Probability and Updating

Part 2: RISK ANALYSIS AS DECISION SUPPORT

6: Example of Risk Analysis and Potential Pitfalls
7: Bayesian Networks and Influence Diagrams
Guest Speaker: Jan Pietzsch
11
13
8: Value of Information from Tests
Preliminary Homework Due
9: Probabilistic risk analysis: the basics
18
Martin Luther King Holiday (no class)
20 10: Functional Block diagrams; event trees
Homework #1 Due
Part 3: PROBABILISTIC RISK ANALYSIS AND SYSTEMS RELIABILITY
13: Safety Optimization under Economic
11: Fault Trees
Constraints & APRAM Model (Appendix 9)
25 12: Systems Evolution: Markov Model
27
14 and Appendix 8: Expert Judgment
Homework #2 Due
16: Value of Human Safety in Government &
Feb 1 Midterm (In Class)
3 Industry Decisions
17: Future risks. The social rates of discount
18: Human and Organizational Factors in
Probabilistic Risk Analysis: The SAM Model.
8
Piper Alpha example
19: Warning systems in risk management

20: Global risk management: Insuring and

improving the system
10
21: A Dynamic Risk Analysis: Reduction of Fire
Risks in Oil Refineries

Part 4: ECONOMICS AND MANAGEMENT OF ENGINEERING RISKS

Guest Speaker: Roger McCarthy
15
Presidents Day Holiday (no class)
17
Homework #3 Due
22 Guest Speaker: Anne Kiremidjian
24 22: Nuclear Power Plants & Fukushima
Part 5: CASES
Feb/
23: Risk Analysis for the Black Tiles of NASA
24: Anesthesia Patient Risk Analysis Study
29
2
Mar
Space Shuttle
Homework #4 Due
7 Guest Speaker: Jim Ellis
9 25: Concluding Remarks
March 14 Final Exam
12:15 p.m.-3:15 [Room TBD]

Additional Information
Studentswhoenrollinthisclasswillberesponsibleforunderstandingtheanswerstothe
followingfrequentlyaskedquestionsandobservingthestatedpolicies,sopleasereadthis
documentcarefully.Theteachingteamwelcomesyouradditionalquestions.
1.

What can I expect to learn in this course? What will I need to do to succeed?
The objective of this course is to present methods and examples of risk analysis,
mostly in the engineering domain, and to present and discuss the conceptual
framework for using these methods in public and private sector decisions.
The scope and concepts of the course have applications in most domains of
engineering mechanical, electrical, aeronautic, civil, etc. which involve decision
making under uncertainty. We will also discuss personal decisions you might make
and a medical case (anesthesia). Particular emphasis will be put on risk to human
safety from the failure of critical engineered facilities and equipment, and on the
philosophical and political issues involved in the risk-benefit trade-offs.
To understand the methods and examples presented, you will need to use the offered
resources (formal class meetings, office hours, course web pages and forum, other online resources, problem sets and solutions, and each other!) to grasp and reinforce this
material. Specifically, your success will depend on:

2.

Attending and participating in class and review sessions.

Completing reading assignments prior to class discussion.
Completing and turning in homework assignments when they are due.
Doing well on the midterm and final exams
Sharing your questions and insights with one another and the teaching team.

Who should take this class?

This course is designed for graduate students. Seniors are welcome provided that they
satisfy the prerequisites and that they are aware of some potential difficulties, for
example, the discussion of open-ended questions.

3.

Are there any prerequisites for this class?

There are three prerequisites and all are important: Engineering Economy (cash flow
analysis and discounting); Statistics and Probability, in particular Bayesian
Probability (at Stanford Stat 116 or MS&E 120/220); and Decision Analysis either
in the form of MS&E 152 or MS&E 252 or an equivalent course. Additionally,
students should be familiar with the basics of Markov Processes and with Convex
Optimization. While risk analysis applications of Markov processes and optimization
will be taught, it will be assumed that students have seen these methods before.

4.

I want to establish a relationship with Professor Pat-Cornell and the teaching team.
Were looking forward to getting to know you, too! Please fill out the online form
(found at http://goo.gl/forms/McAESTkAfb) in order to tell us a little about
yourself. We want to learn your name as soon as possible, and for this purpose, we
ask that you to introduce yourself when you speak in class and to sit in the same place
for each class. Part of the course is a dialog about risk-related issues; your active
participation will help us get to know you. You may also participate via the
Coursework discussion forum, but for in-residence students this does not substitute
for regular attendance.

5.

Do I need to attend every class?

Yes, attendance in mandatory, unless you need medical attention and have to
stay at home. In all other cases, we expect you to attend all classes. We are covering
a large amount of material in this course and you are responsible for all the material
covered in class discussions. If you must miss one class, please let us know in
advance. If you think that it will be necessary for you to miss more than a couple of
classes, we recommend that you register for another course. We will do our best to
start and end each class on time and ask you to do the same. In-residence students
will not be granted access to SCPD-recorded lectures except as required by the
Stanford Office of Accessible Education / Disability Resource Center.

6.

When are homework assignments due? Can I turn in my homework late? How will
homeworks be graded? What resources may I use without violating the Stanford
Honor Code?
Homework must be handed in class on the due date shown in the course calendar.
Homework submitted after solutions are posted on the website will not be accepted.
Complete solutions to all problems will be posted on Coursework.
You are allowed to discuss the homework problems with classmates. However, you
must solve the problems and write up your own answers alone. Use of other peoples
work when not allowed, or failing to note your sources when allowed, is an act of
plagiarism and a violation of the Stanford Honor Code. It robs you of the opportunity
to practice and learn on carefully crafted problems, and does nothing to prepare you
for exams.

7.

What should I do if I have a medical condition?

Stanfords policy is that all accommodations for disabilities (both temporary and
permanent) be coordinated through the Office of Accessible Education / Disability
Resource Center. Their website is http://studentaffairs.stanford.edu/oae . Please speak
to them, and provide us with whatever documentation they prepare for you. We will
not ask any questions. Please do not ask us for accommodations for a medical

condition without having first spoken to the OAE/DRC. We are not medical
professionals, and we have no basis on which to evaluate your request.
8.

I have a schedule conflict. May I write the midterm (or final) at a different time or
location?
All students are required to take the exams on the designated date, time, and
location. (SCPD students have specially designated exam locations.) There is no
alternative final exam. The date and time for the final, Monday, March 14, have been
scheduled by the university, and we have no ability to change it. If you have an
emergency and are unable to attend an exam, please talk to one of the CAs
immediately, and we will try to work with you. Under no circumstances do travel
plans for spring break constitute an emergency. Do not make travel plans before the
final.

9.

What if I have a question that was not answered in the class or review session? What
if I have a question of a personal nature?
We want to hear your questions in class. However, if later you have general-interest
questions about course material or administration, they should be posed through the
discussion forum on Coursework (if you are wondering about something, chances are
good that other students are too). Students can submit questions to the forum at any
time and are welcome to contribute answers as well. Therefore, email to the course
assistants should be limited to personal matters related to the course. The course
assistants will typically provide feedback within one working day.
The teaching team will be glad to discuss the course or issues of interest to you on an
individual basis. Please come see us during office hours, contact us via email, or
make an appointment to see Prof. Pat-Cornell through her assistant Lissa Zelaya if
you cannot make her office hours.

10.

What opportunities are there to continue my studies in engineering risk analysis?

MS&E 250B is a project course based on MS&E250A that will take place in the
Spring quarter. The students work in groups of 4. A number of topics and clients
will be suggested by us (examples of past projects: risk analysis for the design of a
communication satellite; risk analysis for the maintenance of the Boeing 737 between
flights, risk analysis for medical procedures at Stanford Hospital, risk analysis and
decision support for a global automobile manufacturer). If you have a special interest,
you may also define a suitable risk analysis project and propose it to us.

11.

What materials or assistance are permitted during the midterm and final exams
without violating the Stanford Honor Code?
All exams must represent your individual work. You may use a standard calculator
and a dictionary, and you may ask clarifying questions of the teaching team members.

No other discussion, reference material or aids of any kind are allowed during the
exams. Please respect your fellow students and avoid scoring penalties by turning in
your completed exams immediately after time is called.
12.

My homework (or exam) was not graded correctly. How do I appeal the score?
Within seven days of a graded exam or homeworks return, a student may resubmit to
the course assistants a document that they feel has been scored inaccurately. A brief
statement explaining the students position should accompany the submission as a
separate document marked Appeal Request. Altering an original submission prior
to resubmission is a violation of the Stanford Honor Code. The teaching team will
carefully review and completely re-grade each resubmitted document.

Preliminary Homework
Due Monday, January 11, 2016
Problem #1
Consider a cash flow stream representing some investment, for example an infrastructure
project that is expected to generate revenue. Given the following time stream of cash flows
(in thousands of dollars) and a 15% annual discount rate:

a) Find the net present value (NPV) in year 0 dollars of the cash flows for years five through
ten, year 5 included.
b) Find the NPV at year zero of all cash flows as a function of the unknown cash flow X in
year four.
c) Assume that the cash flow X in year four is $20K. The project manager has an alternative
investment, which is an annuity consisting of uniform cash inflow of Y thousand dollars
per year starting in year 1 and ending in year 10. Note that if the manager chooses the
annuity, no cash flow will be received at year 0. What value of Y would make him
indifferent between the proposed project and the annuity?

Problem #2
A discrete-time Markov process has 3 states, which are numbered 1, 2 and 3. The transition
probabilities are denoted pnm and represent the probability that the process transitions from
state n to state m in any one period. Assume that the process with the following transition
probabilities:
p11= 0.27
p12= 0.23
p13= 0.5
p21= 0.8
p22= 0.13
p23= 0.07
p31= 0.13
p32= 0.2
p33= 0.67
a)

Draw a state diagram representing this process.

b)

If the process begins in state 2 with certainty, what is the probability that it is in state 1
after 3 periods? (You may perform the computation with the aid of a graphing
calculator or computer software, but please ensure you understand the computation
method.)

c)

Write the global balance conditions, and use them to find the steady state probabilities

d)

If the transition probabilities were changed to p31=p32=0, p33=1, how would this change
the steady state solution? Explain.

Problem #3
a)

Suppose that X and Y are jointly discrete random variables with probability mass
function

xy
for x 1, 2 and y 2, 3, 4

p( x, y) 27
0 otherwise

Find px(x) and py(y).

Are the random variables X and Y independent?

b)

Let the following density functions be given:

4 2
x ,
f X ( x) 3
0,

0<x<1
otherwise

(Probability density function)

1
2,
f Y | X ( y | x) 2 x
0,

0 < y < 2x2

otherwise

(Conditional density function)

Find: f Y ( y ) and f X |Y ( x | y )

Problem #4
Your company uses 90,000 units of a certain type of memory chip each year. Companies L,
M, and N each supply you with 30,000 chips per year. You test each chip upon arrival, and
have found the probabilities that a chip will fail are 1.5 x 10 -3, 1.5 x 10-4, and 1.2 x 10-2 for
each company's product, respectively.
a) What is the probability that a given chip will fail if you don't know from which vendor it
came?
b) Given that a chip has failed, what is the probability that it came from Company L?
c) If you have two chips, and know from which vendor each came, are the failures of the
chips independent events?
d) If you have two chips and know that they came from the same vendor but don't know
which one, are the failures independent?

NOTE: Do not "guess". Your answer must be based on probability computation whenever
necessary. State any assumptions you made in arriving at your answer.

10

Problem #5
South Bay Air (SBA), a startup passenger airline, has an aircraft fleet that provides a capacity
of carrying 15,000 passengers per year. In making a three-year forecast, the marketing
manager estimated that yearly demand will be as follows:
Passenger/Year
Probability
Revenue/Year
Low
15,000
0.4
$250K
Medium
20,000
0.5
$330K
High
32,000
0.1
$500K
As VP-Operations, you are convinced that SBA is a highly efficient airline and that capacity
can only be increased by acquiring new aircraft. You must make a presentation of a threeyear plan (from 2011 through 2013) to the Board of Directors. Which of the following
capacity options will you recommend?
(a) Do nothing.
(b) Lease exactly one airplane from East Bay Air, available as soon as you pay the lease.
Lease fee for two years is $100K (you can only lease the plane for two years) and a lease
can be entered only January 1, 2011 or January 1, 2012. This provides additional
capacity for 5,000 passengers per year. You must decide now (i.e., before seeing the 2011
demand) if you will lease a plane.
(c) Buy a new airplane, available immediately (i.e. 1/1/11. A plane cannot be bought after
this date). Cost is $200K, payable on April 1, 2011. This provides additional capacity of
17,000 passengers per year.
This forecast is for three years only. At the end of this time, the state of California will
revoke your license and nationalize your equipment (i.e. no salvage value). Passenger
demand will be the same for the first two years (e.g., if it is medium in 2011, it will be
medium in 2012), but it may change again in 2013, independently of previous years.
ASSUME:

The discount rate is 10% compounded monthly. The 10% is the nominal
yearly rate, not the effective yearly rate.
All calculations are in dollars of 1/1/2010.
Ignore operating costs.
Revenues occur at end of year.
Future revenues and costs have not been discounted.
SBA is an expected value decision maker.

Problem #6
Describe an example of a risk analysis problem from a commercial, financial, technological,
or medical field. What are some of the uncertainties that need to be modeled? What is the
risk of concern? What might the policy or decision ramifications be for an analysis?

11