יום שלישי, 8/5/2012
13:00-18:30אולם 056, בנין שאפל
הפקולטה להנדסה, אוניברסיטת תל אביב
תכנון ואופטימיזיציה של תהליכים עם יישומים הנדסיים
Process Design and Optimization for Engineering Applications
A special workshop by Geoff Vining, Virginia Tech
יום עיון עם מחבר רבי מכר מפורסמים בסטטיסטיקה תעשייתית:
A half day workshop delivered by the author of famous industrial statistics bestsellers:
- Vining, G. Statistical Methods for Engineers. Belmont, Ca., Duxbury Press.
- Park, S. and Vining, G. Statistical Monitoring and Optimization for Process Control. M. Dekker
- Montgomery, D., Peck, E., and Vining, G. Introduction to Linear Regression Analysis, J. Wiley.
- Myers, R., Montgomery, D., and Vining, G. Generalized Linear Models, New York: J. Wiley.
In this workshop you will learn:
1) Why increasing process temperature reduces particle size?
2) What is sequential experimentation?
3) How to optimize a wafer fabrication process using response surface experiments?
4) Can reliability of electronic components be evaluated with experimental design?
Target audience: Industrial engineers, chemical engineers, process engineers, quality engineers, electronic engineers, reliability engineers and industrial statisticians.
This workshop is sponsored by the Israel Statistical Association who will issue a special completion certificate signed by Professor Vining to preregistered participants.
רישום מראש מקנה זכות לתעודת השתתפות של האיגוד הישראלי לסטטיסטיקה עם חתימתו של פרופ' וינינג.
ההשתתפות כרוכה בתשלום של 150 ₪ (120 ₪ לחברי האיגוד הישראלי לסטטיסטיקה).
הכניסה לסטודנטים חינם, עם רישום מוקדם והצגת תעודה בתוקף.
רישום ותשלום ע"י שליחת דוא"ל עם הכותרתVining אל:
Yisrael Parmet (email@example.com)
קבלה תינתן ביום האירוע על ידי גזבר האיגוד
Industrial statisticians have been applying successfully basic experimental designs and analyses to improve products and processes for many years through response surface methodology (RSM), which is a sequential approach to experimentation. RSM originated in the chemical process industries as an effective and efficient means to optimize both new and existing processes and products. Over the years, RSM has become a cornerstone for process improvement strategies, including Six Sigma. This course uses real examples to illustrate the basic approaches for process design and optimization. Throughout the course, we use a simulated distillation column to do actual experiments in real-time.
The first example uses a fractional factorial experiment to understand how to increase production of peanut oil. A chemical engineer looked at the impact of process pressure, process temperature, moisture in the peanuts, the carbon dioxide flow rate (used to extract the oil from the peanuts), and the peanut particle size on the amount of oil extracted. She learned that to increase the amount of oil extracted, she needed to increase the process temperature and reduce the peanut particle size.
The second example illustrates sequential experimentation to improve the yield of synthetic jojoba oil, used in food manufacture. A group of Spanish chemical engineers sought to find the conditions for reaction temperature, initial amount of catalyst, and reaction pressure on the yield of the oil. The actual experimentation proceeded in two phases. The first used a full two-level factorial design plus some “center runs.” They discovered a problem with curvature. As a result, they augmented their original experiment to estimate a full second-order model. They then determined the conditions to maximize the yield.
The third example shows how SEMATECH used a second-order response surface experiment to improve a silicon wafer process. A group of material scientists sought to maximize the deposition rate of tungsten film on a silicon wafer by changing the chamber pressure and the chamber gas composition. They had prior information to suggest an appropriate experimental region to apply a second-order response surface experiment. They then conducted the experiment and determined the optimal conditions.
A fourth example examines how to apply these techniques to model the reliability of an electronic component. An electrical engineer studied the impact of operational temperature and voltage on the lifetime of this part. This workshop shows an appropriate analysis of this experiment using standard statistical software.
The workshop emphasizes the analysis of real engineering examples and the use of statistical software, such as Minitab and SAS-JMP. This workshop will also review the basics of designed experiments, especially factorial designs, fractional factorial designs, and normal probability plots for analyzing unreplicated experiments.
Workshop Outline (13:00 – 18:30)
- Overview of RSM
- Historical Perspective
- Concept of Sequential Experimentation
- Common Models
- Review of and Designs
- Center Runs and Lack of Fit
- Path of Steepest Ascent
- Augmenting and Designs
- Central Composite Designs
- Second-Order Experiments
- More on the Central Composite Design
- Analysis of Second-Order Experiments
- Box-Behnken Design
- Single Response
- Multiple Responses
- Applying Response Surface Techniques to Reliability Data