Im Rahmen des 3rd European DOE User Meetings werden insgesamt 3 optionale Workshops angeboten.

Termin

Das Workshop "DOE and RSM Simplified" by Mark Anderson, Stat-Ease, Inc. ist für den 31. Mai 2010, also einen Tag vor dem eigentlichen User Meeting terminiert.
 

Sprecher

Mark J. Anderson is a principal and general manager of Stat-Ease, Inc. Prior to joining the firm, he spearheaded an award-winning quality improvement program for an international manufacturer, generating millions of dollars in profit. Mark offers a diverse array of experience in process development, quality assurance, marketing, purchasing, and general management.
 

Agenda
 

 
Abstract
 

This workshop offers an overview of design of experiments (DOE) and response surface methods (RSM). It illustrates an array of practical statistical tools for design and analysis of experiments aimed at breakthrough improvement and process optimization. Although this workshop is not a substitute for hands-on computer-intensive training, it will give participants a foundation for upgrading their DOE/RSM skills to a superior level.

The morning session introduces basic concepts of DOE and then explores full factorials, interactions, fractional factorials, and aliasing. Several informative case studies will be presented in detail. The afternoon session progresses into response surface methods (RSM). Once again, the material will be presented in case-study format that translates these statistical tools most effectively. During this session, participants will learn to appreciate all that RSM can do to find their sweet spot -- the ideal settings for the critical process factors.

Im Rahmen des 3rd European DOE User Meetings werden insgesamt 3 optionale Workshops angeboten.

Termin

Das Workshop "Formulations Simplified" by Pat Whitcomb, Stat-Ease, Inc. ist für den 31. Mai 2010, also einen Tag vor dem eigentlichen User Meeting terminiert.
 

Sprecher

Pat J. Whitcomb is the founding principal and president of Stat-Ease, Inc.. He co-authored Design-Ease software and Design-Expert software. He provided consulting on the application of design of experiments (DOE) and other statistical methods for mor than two decades. In addition, he is co-author of the books e.g. "DOE Simplified: Practical Tools for Effective Experimentation".
 

Agenda
 

 
Abstract
 

If you do product formulation, then standard factorial and response surface designs just don't work. You need to use mixture designs to ensure success. The one-day Formulations Simplified workshop teaches you how to effectively create mixture designs for a variety of problems. Develop statistical models of product performance. Then use response surface methods to identify the "sweet spot" where the optimum tradeoffs among multiple responses occur.

Ingredients for Efficient Experimentation

See the outline below for a list of topics you'll be introduced to in the Formulations Simplified workshop:

• What makes a mixture?
• Mixture (Scheffé) polynomials
• Simplex lattice designs
• Synthetic medium case study

• Detergent formulation case study
  • Coding: Actual, Real, L_Pseudo
• Optimization of multiple responses
  • Numerical (desirability function)
  • Graphical (overlay plot)

• Sizing for precision
• Constrained mixtures, extreme vertices
  • Shampoo case study
• Optimal point selection
  • Flare case study

• Multicomponent constraints
  • Fruit punch case study
• Mixture amount experiments
  • Ibuprofen case study
• DOE support

Im Rahmen des 3rd European DOE User Meetings werden insgesamt 3 optionale Workshops angeboten.

Termin

Das Workshop "Multivariate Data Analysis in a DOE context" by Ivan Langhans, CQ Consultancy ist für den 03.Juni 2010, also einen Tag nach dem eigentlichen User Meeting terminiert.
 

Sprecher

Ivan Langhans, PhD. Chemestry, M.Sc. Statistics, founder of CQ-Consultancy, the Belgian Chemometrics Society and B-Enbis, part-time researcher at the Faculty of Business and Economics of the KULeuven. His research interests ar model selection and model validation.
 

Agenda
 

 
Abstract
 

Multivariate methods such as Principal Component Analysis (PCA) have been developed to deal with correlations between variables. In this workshop a brief introduction to Principal Component Analysis will be given, with two illustrations of its use and power in a DOE context.

You may need to test different solvents, additives, catalysts, system configurations, …. In that case you’re dealing with so-called categorical variables. Since categorical variables with n levels can be considered as n-1 variables, the number of experimental runs needed in a design increases rapidly with the number of levels to be tested, and often we can’t test them all.

A typical example is the choice of solvents to be used. For this example a first step in Principal Properties Design is to describe the solvents by a set of quantitative properties that are related to what makes them different (e.g. polarity, mass, di-electric constant, …), which results in a set of ten to twenty highly correlated variables. The next step is then to perform a PCA and select a representative set of solvents based on the principal components that adequately describe the set of descriptors, i.e. the principal properties.

With Principal Properties Design the problem of multi-level categorical factors can be circumvented, and on top, we even get information about items / levels that have not been explicitly tested.

Text-book examples of design often consider studies with just one or two responses. In real life you may have dozens. Again PCA can reveal the correlation between all those responses which helps identifying outlying values and prove unattainable targets (e.g. low value for response Y1 combined with high values for Y2 when they exhibit strong positive correlation).