Design of Experiments - 230421

Question 1

Steps for DOE planning / Statistische Versuchsplannung

Answer 1

1. Describe/expertal knowledge: fact sheet, collect expert knowledge
2. Select factors and responses: Which ones could have an effect? Previous technical documents. Measurements Systems Analysis.
3. Determine design space: Parameters and ranges of values
4. Choose a DOE design:

Question 2

What is the Measurements Systems Analysis?

Answer 2

Provides info about the exactness of adjusted settings. Gauge calibrated and gauge R&R
Hear the signals of the process much easier than an unstable one.

Question 3

Fact sheet

Answer 3

• Collection of data, after sorting and structuring.
• Problem Statement / target as specific as possible: 99% of all parts have a tensile stress of 60N
• Process description: chronological processes, cause and effect diagram (fishbone diagram, probably not a very professional diagram)

Question 4

factors

Answer 4

• data type: adjustable, text (limited number of values)
• Adjustability: adjustable (easily changeable), hard-to-change, not adjustable-but measurable, not adj not measure
• discard as many factors as possible.

Question 5

Design space

Answer 5

region of experiments, always within the technical limits

Question 6

What are the different available designs besides two factorial design?

Answer 6

Factorial designs: full factorial design,fractional factorial: 2^(k-p)
Response surface designs: Central composite design
Mixture designs: Simplex centroid, simplex lattice

Question 7

Center points

Answer 7

An intermediate point between the -1 and +1 values, allows a non linear relation

Question 8

Factorial design

Answer 8

2-50 factors, could provide a great improvement on the response. Selects vital factors out of many, outcome is a hyperplane n-dimensional plane

Question 9

advantages

Answer 9

simple setup
No knowledge about
Evaluation of all main effects

Question 10

disadvantages

Answer 10

number of tests required grow exponentially (only makes sense until k=5)
Equations might be challenging
After three factors, no physical interpretation

Question 11

Fractional factorial designs

Answer 11

Full: some experiments might be irrelevant for the DOE, but nevertheless they must be carried away
Fractional: 2^(k-p),