IE 511 Experimental Design in Engineering
PSU, Fall 2021
Please note that the specifics of this Course Syllabus are subject to change. I will notify
students of any changes and students will be responsible for abiding by them. Even if you
print this syllabus, please check the online version often.
Instructor: Professor Enrique del Castillo, Dept. of Industrial and Manufacturing Engineer-
ing, email: please used CANVAS e-mail. Lab web page: [Link]
engineeringstatistics/
Class meetings: T Th 10:35am to 11:50am, via zoom in “synchronous” (real time) mode.
Note all times are Eastern US times. Lectures will be recorded. Please visit the remote
learning help page at [Link] for Frequently Asked Questions
and helpful links for remote learning assistance. You must be able to access the CAN-
VAS class organization system and be able to use the Zoom app for remote video/audio.
For an introduction on how to use Zoom, please see [Link]
participating-zoom-meeting-quick-start-guide For computer requirements you must
meet in order to access Canvas, see [Link]
[Link].
Office hours: Tuesdays, 12:00 noon to 1pm by zoom. Office hours are for you to ask ques-
tions about the course, not to resolve point grading disputes (see below).
Course objective: this is a first year, graduate-level course in designing experiments and an-
alyzing data collected from an experiment. The course deals with the types of experiments
that are frequently conducted in industrial settings. Both design and statistical analysis
issues are discussed and their relation to machine learning and “big data” is emphasized.
Please note this is not a replacement for a generic “Statistics” class. You must know Statis-
tics before attending this class.
Prerequisites: One full semester undergraduate course in Probability and one full
semester undergraduate course in Statistics. A list of concepts you should be familiar
with is given below. If you have not taken the prerequisites or are not familiar with the
prerequisite material, you cannot take this class.
Course homepage: I will use ”CANVAS” for class support and for hosting the video links. I
will be posting class material on the course page regularly (under ”Files”), so please check it
daily. To contact the instructor by e-mail please use CANVAS email so that I receive your
email.
Textbook: D.C. Montgomery Design and Analysis of Experiments, 10th ed., Wiley, 2020.
9th, 8th and 7th editions are OK too. This is a good but almost undergraduate-level book,
so considerable additional material will be covered in class or distributed via CANVAS.
1
Additional textbook material is available (for free) from the 9th edition at: [Link]
[Link]/he-bcs/Books?action=index&itemId=1119320933&bcsId=10790
Other useful references are:
• Del Castillo, E., Process Optimization, a Statistical Approach, Springer, 2007 (available
through SpringerLink in the PSU Library).
• Wu, C.F.J., and Hamada, M. Experiments, Planning, Analysis, and Optimization.
Wiley, 2009.
Grading System
• Final project report (25%), due December 10 via CANVAS. One page project pro-
posals due October 22. More information about the project, which you should keep
present, is at the end of this handout.
• Preliminary Exam I, September 23 (during class time, no class that day), 25%
• Preliminary Exam II, November 4 (during class time, no class that day), 25%
• Final Exam, (to be scheduled by the university) (25%)
Grading scale: 90-100% → A; 85-90% → A-; 80-85% → B+; 75-80% → B; 70-75% → B-;
65-70% → C+; 60-65% → C; 55-60% → D; 0-55% → F (no ’curving’)
Point questions or disputes. These will not be discussed during office hours. If you have
questions about why you did not receive full points for any exam questions, you have one
week from the date the graded exam is returned to email an explanation of any issues to the
professor. Include your exam in question, and a clear explanation of why you think there is
an issue.
Homework: some exercises will be suggested in class, but will not be collected/graded.
Exams: In both preliminary exams and in the final exam you will have a given time, TBA,
to upload your solutions from the time the exam becomes available to download. No late
submissions will be allowed.
Class format and note taking: The class is offered in a “remote synchronous” (live) format.
I will be using a tablet to write on a white board as I speak and occasionally show some
graphs and figures. Any handouts will be posted via CANVAS. I suggest you take notes,
so that you have at the end a more permanent record of the material you learned. The
synchronous mode of instruction allows for questions, so feel free to ask questions when you
do not understand something. To access the class, login to the CANVAS course web page,
and click on ”Zoom” in the left navigation panel in Canvas.
2
Emailing through Canvas - Students are expected to use Canvas for all course email com-
munication with the professor and students are expected to be checking their email and
notifications at least daily.
Videoconferencing etiquette. Please keep your microphones off when not speaking so that
you don’t inadvertently create disruptions. Questions are welcome, but please turn off your
microphone when you are finished asking a question.
Statistical software: Be advised you will not be allowed to use software during examinations.
However, modern practice of Statistical Methods in industry heavily relies on statistical
software. Therefore, in class I will sometimes show output from several different software
packages. The textbook uses MINITAB, a menu-based software widely available at PSU and
SAS JMP, also available at PSU. I will use the R statistical system in some sections, which
is a modern software wildly used today fro data analysis.
Students with disabilities. Penn State welcomes students with disabilities into the Univer-
sity’s educational programs. If you have a disability-related need for reasonable academic
adjustments in this course, contact the Office for Disability Services (ODS) at 814-863-1807
(V/TTY). For further information regarding ODS, please visit the Office for Disability Ser-
vices Web site at [Link]
In order to receive consideration for course accommodations, you must contact ODS and
provide documentation (see the guidelines at [Link]
documentation-guidelines). If the documentation supports the need for academic ad-
justments, ODS will provide a letter identifying appropriate academic adjustments. Please
share this letter and discuss the adjustments with your instructor as early in the course as
possible. You must contact ODS and request academic adjustment letters at the beginning
of each semester.
Tentative syllabus and recommended reading
approximate
# of lectures Topic(s) Reading (Book sections)
1 Introduction: DOE and the “big data” world Ch. 1 (all)
2 Review of basic statistical concepts 2.1 to 2.5
4 Experiments with a single factor 3-1 to 3.6, 3.8
1 Randomized block and latin square designs 4.1, 4.2
4 Introduction to factorial experiments 5.1 to 5.5
k
4 2 factorial experiments Ch. 6 (all except 6.7)
1-2 Blocking and confounding in 2k experiments. 7.1 to 7.7
4 2k−p fractional factorial experiments. 8.1 to 8.7
3-4 Regression models 10.1 to 10.7
1 Intro. to process optimization and RSM 11.1 and 11.3
2-3 Experiments with random factors 13.1 to 13.5
2-3 Nested and Split-Plot experiments 14.1 to 14.5.1
3
Prerequisite concepts
I will assume you know the following material. If you are not familiar with these notions
it means you do not have the prerequisites to take the course:
• Random variables
– formal definition of a random variable
– Functions of random variables
– Expectation, Variance, and Covariance operators acting on random variables,
properties of these operations
• Random sampling: formal definition of a random sample. ”IID” (or iid) random
variables
• Properties of point estimators (bias, variance, mean square error)
• Sampling distributions (distributions that originate from sampling experiments):
– Normal, standard normal, standarization
– Student’s t
– χ2
– F
(Here I am not interested in the use of ”tables”, but that instead you know the
relations between these distributions (the normal and the t, the normal, the t, and
the χ2 , and the χ2 and the F), when these distributions are used (i.e., in which
types of problems), and how they arise in probability and statistics problems).
• Use of sampling distributions to build confidence interval estimators and test hypothe-
ses for:
– mean and variance of a normal populations
– differences between means and variances from two different populations.
• Basic matrix manipulations: inverses, transposes, matrix multiplication, solution of a
linear system of equations, linear independent vectors.
4
Term Project
The practice of experimentation can only be learned by doing an experiment. In this
project you will plan, design, conduct, and analyze the data from an original experiment
created by you using DOE techniques. The project is to be conducted strictly individually.
The context of the project is limited only by your imagination. Please note that analyzing
“historical data” from an experiment conducted in the past by you or somebody else is not
allowed, you need to design and run a new experiment yourself. It is preferred, but not
mandatory, that you conduct an experiment related to your own research or related to some
other class or lab that you currently take. If this is not possible, conduct a “household”
experiment (e.g., the kitchen and some home appliances provide a good “laboratory” to
conduct non-trivial experiments with some “equipment”; try to be creative). To get a feeling
of the type of projects that are acceptable, sample projects can be found in the additional
text material web site, but you must be original.
Project proposals are due October 22, 5pm ET, uploaded to CANVAS (no email
submissions). This should be a one-page, typed description of the experiment you wish
to conduct, listing the following:
1. the goal of the experiment (what you want to investigate);
2. the experimental units in your experiment;
3. the controllable factors you wish to vary, the response(s) of interest you will investigate
and how you will measure the response(s).
4. any other factor you will keep constant and any nuisance factor you may wish to block.
The purpose of the proposals is to assure that all experiments will result in a similar ef-
fort and educational experience and to avoid problems later on. Proposals will be graded
acceptable/non-acceptable, and you will have up to 2 chances to present an acceptable
proposal, otherwise you will fail the project mark. Therefore, please spend time thinking
about your project and writing a good proposal.
Major requirements of the project are:
1. the experiment must involve at least three significant design factors (after screening),
so the number of initial (potential) factors affecting the response(s) must be larger
than this (I suggest at least twice this, so start with at least 6 factors). All the factors
must have a real potential of being active, at least at first sight. It is precisely the
purpose of the experiment to determine their effects.
2. The experiment should not be trivial: a proposal of an experimental situation where it
is obvious what will happen before hand is not acceptable (e.g., ”We wish to investigate
the effect of water temperature and boiling time on the hardness of an egg”). Therefore,
you should spend considerable time selecting the project theme.
5
3. Without exceptions, the experiment must be a physical experiment, conducted with
some machine or equipment, where there are some physical experimental units and
therefore it is necessary to randomize the experiment, as opposed to an experiment
conducted on a computer/internet, I-phone, I-pad or similar; i.e., simulations or emu-
lators of some real-life system, are not allowed. Hint: an experiment is “physical” if it
consists of performing tests on experimental units (specimens) that are physical,
not virtual or digital.
More information about the project will be provided after the project proposals are due.