ETLS 701 Design of Experiments (DOE)

Instructor:

Perry Parendo, P.E

perry_par2002@yahoo.com

Time:

Section 09, Wednesday, 6:00 – 9:00 p.m.

Location:

St. Paul Campus, OSS LL10

Telephone:

651-962-5750 (Engineering Office)

Required Text:

Anderson, Whitcomb. DOE Simplified, Stat-Ease Corporation,
Design-Ease Version 7 software comes with text.

Optional Text:

Cornell, How to Apply Response Surface Methodology

Cornell, How to Run Mixture Experiments for Product Quality

Course Description:

This course provides the student with a set of skills to improve products and processes already in manufacturing as well as to develop products and processes in the development stages. The definition of DOE promoted is “a tool to assist in the process of understanding a system.” There will be discussion of how DOE fits into the overall product lifecycle and where it applies in the area of testing. Tools covered include full and fractional factorials, central composite, Box-Behnken, Taguchi, Evolutionary Operation and the method of steepest ascent. Theoretical statistics understanding is assumed prior the course. A standard, simple process will be presented which allows for improved communication and user confidence in using the tool set. The primary objective is to assist the student with implementing the skills learned during the course. This is an application-orientated course that includes case studies, team projects, student presentations and reports, guest lecturers and use of computational software. A quick statistical overview will be provided in the class as a refresher, but is not intended to cover the subjects in depth to students new to the subject. It is recommended students review all of the topics prior to starting the class.
Prerequisite: ETLS 506 Statistical Methods for Manufacturing Quality

Background Required:

The underlying statistics covered in ETLS 506 (previously MMSE 615) are a critical foundation for the material presented in the DOE class. The minimum background needs to include an understanding of: statistical symbology; normal distribution; ANOVA; and z, t, and F tests. General understanding of alpha and beta errors, flow charts, Pareto charts, Gage R&R studies, cause and effect diagrams and SPC is nice but not required. If you have equivalent background, the instructor can grant a written waiver

Undergraduate students: Undergraduates may apply for the course but should contact the instructor ahead of time to discuss expectations.

Note:If you believe you already are proficient in this material, contact your advisor to discuss alternatives.

Course Objectives:

This course will provide the student with a set of skills that can be used to improve products and processes. These skills may be applied in manufacturing or be used to develop products and processes during the development stages. A definition of DOE is “a tool to assist in the process of understanding a system”. We will discuss how DOE fits into the overall product life cycle and in particularly where it applies and does not apply in the area of testing. Tools covered include full and fractional factorials, central composite, Box-Behnken, Taguchi, Evolutionary Operation and the method of steepest ascent.

Learning Outcomes:

Upon successful completion of the course, the student will be able to:

1. apply the project set up and approach assessment methods for DOE execution (P) (ABET SE3, SE9, SE10, MS1, MS4, MS5)
2. comprehend that different DOE approaches are valid for solving the same problem (HW, E) (SE8, MS1)
3. demonstrate ability to identify an appropriate DOE project (E, P) (ABET SE12, MS1, MS5)
4. indicate awareness of and ability to interpret advanced DOE tools and topics including optimization (HW) (SE15, MS1)
5. demonstrate a thorough working knowledge of factorial DOE trade offs and analysis (HW, E, P) (ABET SE3, SE 11, SE12, MS1, MS4)
6. comprehend and internalize the DOE process (E, P) (ABET SE3, MS5)
7. fluently express DOE information to peers in written and presentation format (R, P) (ABET SE9, SE12, MS2)

Course Methodology:

The primary objective is to implement the skills learned as a part of the course. Because of the availability of inexpensive PC software to perform the statistical calculations, the mathematical focus will be on the assumptions made during the analysis and how to check those assumptions. A standardized, simple process will be presented which allows for improved communication and user confidence in using the tool set. Projects, reports and practice examples of real applications will allow the student to apply the tools and the process to real life problems and gain practical experience. This is not an analysis class or a software class – it is a thinking class.

Major
Assignments:

1.     Project (P) - An actual experiment will be conducted by each student on a topic which they select. Ideally these projects will be from actual work needs. Past projects will be provided. The reviews during class will allow for coaching at key points in the experimental process.

Definition - first 3 steps (hand in and discuss)        Due 5

Status/Approach - up to ready-to-test (discuss)     Due 10

Final Presentation - (hand in a report and discuss) Due 14

2.     DOE Topic Report (R) - A DOE topic of interest to the student will be selected to explore in more depth than is covered in class. Past topics will be provided.

Outline                                                              Due 6

Report and presentation                                   Due 10

3.     Exams (E) - These will be problems from actual experiments which represent reality closer than typical text book problems.

        Mid-term, take home 7                              Due 8

        Final, take home 13                                  Due 14

4.     Weekly homework (HW) - Short questions and simple examples from a variety of text books. Solutions will be posted on BlackBoard after class. If you desire extra problems for practice, notify the instructor.

Grading Policy:

Project Presentation and Report - 50%
Topic Report - 18%
Exams - 25%
Homework - 7% (HW #6 is in class and worth 2%)

A: Student shows insight and creativity in implementing DOE understanding. Student has the ability to be a DOE project lead.
B: Student could lead a DOE project with some coaching. Student understands concepts but uncomfortable in implementing process.
C: Student met the minimum expectations. Student would be a supportive team member of a project.

All work shall be turned in at the start of the class period. Delays due to travel or other business constraints shall be coordinated in advance.  Delays due to illness shall be submitted [homework only] by noon the following day.

Academic Integrity:

All students are expected to understand and follow the University of St Thomas policies on Academic Integrity. These are described at: www.stthomas.edu/engineering/graduate/policies

Exams: Exams are one of the instruments used to evaluate the knowledge gained by an individual student of the class subject matter, and the progress towards meeting the outcomes of the class and the degree.  To this end all exams (in class or take home) are intended to represent the effort of the individual and not a group effort unless specifically stated otherwise.

Attendance
Policy:

Students are expected to attend all class sessions. Circumstances which prevent attendance will be honored up to two instances. Absences in excess of two times may result in a lower grade for the course. Contact the instructor when a special situation arises. All absences require that the instructor be informed in advance.

Students with Disabilities

Qualified students with documented disabilities who may need classroom accommodations should make an appointment with the Enhancement Program – Disability Services office during the first two weeks of the semester.  Appointments can be made by calling 651-962-6315 or in person in O’Shaughnessy Educational Center, room 119.

Instructor Background:

Perry helps companies achieve product and process breakthroughs using Design of Experiments (DOE). Perry leverages extensive DOE experience in manufacturing and product development to bring you an innovative yet effective methodology for all developers desiring a quantum-leap improvement in product or process effectiveness.

Perry began developing and seeing results from his DOE techniques at the GM Research Labs in 1986. He has refined and tested them in various environments, including automotive, aerospace instrumentation, armament technology and medical products in roles ranging from R&D, to design and production. His unique insight into how DOE is best used has saved time and money while solving problems in all of these industries.

Parendo has been grooming a new generation of experts at the University of St. Thomas through his teaching since 1996. He has spoken to domestic and international audiences on the use of DOE techniques, and consulted in a wide range of industries. He received his Mechanical Engineering degree from the University of Minnesota.

Design Of Experiments In Industrial Testing
and
How It Relates To Your Business

Movement in Testing

Testing has historically been able to obtain the customer system performances desired with reasonable efforts. Today, however, customers require faster changes in technology and requirements are getting more difficult to meet. Competitive cost pressures have made the achievement of goals an even tougher task. In some cases, customers have requested and in some cases mandated tools for companies to "be more efficient and organized in testing". These facts create the need for organized and effective testing methods to be incorporated in all phases of the development process.

A technique is available called Design Of Experiments (DOE) which assists in accomplishing these goals. It uses a statistically based methodology to efficiently plan and execute tests to achieve the desired performance levels. DOE has historically been used in process industries as well as in manufacturing environments. A strong movement has recently begun applying DOE in the area of research and development.

How DOE Works

Design Of Experiments uses work group experience to generate test responses and key variables as is current practice. By knowing the number of variables and a test budget, an efficient and effective testing matrix can be determined. Testing the variable combinations defined in the matrix allows the tester to determine many pieces of information. For example, a DOE test can identify key variables and key interactions between variables, magnitude of effect of key variables, values for optimum performance and tolerance limits or operating ranges for acceptable operation. In addition to these capabilities which are obviously part of design, it can also help in improving product reliability, developing manufacturing processes, improving process capability (decreasing variability) assisting in manufacturing problem solving (SPC) and understanding field failures. It is important to note that DOE not only can improve products and processes but can also lead to optimization and tolerance design.

The DOE process is best performed in a team environment and may include input from engineering, customers, vendors, operators/technicians and management. With the up-front agreement by all team members on variables, test response(s), desired output and project constraints, the DOE process has a high success rate for buy-in.

The engineer or scientist is very involved in the DOE process. If the technical expert is taken out of the loop, and the statistician or DOE expert directs the testing, failure of the DOE process is likely to occur. The technical expert can contribute experience, subject matter knowledge and intuition to guide testing and ensure test results are interpreted properly. Use of a technical expert and an experienced DOE tester is important for success in DOE testing.

Why use DOE at Your Company?

Many areas exist to use DOE tools. It can be used in trade studies, simulations, development testing, manufacturing and analyzing field failures. We can use the tools ourselves as well as assist sub-contracts in using it to optimize and understand subsystems for which they are responsible. This is an opportunity for your company to be a leader in the movement of using DOE in the research and development area.

DOE has proven to be an excellent tool to reduce test time and provides the ability to reach higher levels of system performance by accounting for interactions and quadratic effects. Compared to traditional test methods, DOE can reach similar objectives yet reduce costs by a minimum of a factor of 2.

DOE can also be used in the area of computer modeling and simulations, which are areas that many companies have increasingly used. This area of testing could include utilizing Finite Element Analysis code, empirical codes, financial models and many others.

Training

A class in DOE methods has been developed at St. Thomas to meet the need created by this movement in testing. The class trains personnel how to perform DOE in the production environment as well as research and development. The class has been enhanced over typical DOE training to include DOE process steps that are non-obvious in the research and development area. The class covers the statistical tools, the process of applying DOE to manufacturing or development areas and assists the students in beginning to use the tools in their own work areas. The class includes practical application examples and has a project from each students own work environment integrated into the class.

Projects done as a part of the class are encouraged to be small and low risk (technically as well as schedule). The selected project can act as a tool to learn about DOE applications within the class participants work area. It is also encouraged when to use and when not to use DOE as a tool. It is best used in areas that need to be improved, optimized or understood. It is important to note that DOE can only assist in getting the most performance that a concept is capable of. If the concept is not capable of achieving the goal, DOE can not "make the concept work".

The class discusses the classical western methods as well as techniques and philosophies from Taguchi and Shainin. The class takes an overview approach to statistics. Manual mathematical calculations are minimized since software is available. The software, which the class participants will use, will be available on the PC network.

Advantages and disadvantages of each tool are stated. Assumptions made with each tool are identified. Potential testing pitfalls are identified and methods to avoid them are discussed. Solutions for recovery from a pitfall are provided in case of occurrence during testing.

Summary

Recognizing a movement in testing toward the use of DOE in research and development, it is rational for any company to create a DOE competency within the engineering ranks. The course relates the tool to application for all companies needs and teaches how to use the tool intellectually compared to a blind, cookbook approach. Skill in DOE methods will support your companies business and technology needs and is a critical competency for your future.

- - Perry Parendo

[ TOP ]