ME 350 – Lecture 14 – DOE Part 1. Design of Experiments at Grainger in reference section covering chapters 17 & 18 Quality Control Gaussian distributions Quality Loss Function Process Control 2 k Factorial Design. Goal of Quality Control.
Design of Experiments
at Grainger in reference section
covering chapters 17 & 18
Strategic view of Quality Design and Improvement:
Control variable 1, x1: extruder temperature
Control variable 2, x2: injection time
Output measurement, y: part weight (goal is 25 oz)
Variance of a system: σ2 =
Tolerance of a system:σ =
A part with a “hole” must match up with another containing a “pin.” The hole and pin tolerances () are 15 mil. Thus the clearance and clearance tolerance is (assume Gaussian distribution):
Same problem as before (tolerances are 15 mil). Thus the clearance (with tolerance) is:
The deviation of a product from its nominal value typically has a similar deviation in the product performance:
Warranties for transmissions with greater variability are more expensive. They tend to follow a:
1) Choose an output region with less variability for a given input, or 2) tighten the input
“You can not manage what you can’t measure”
“If you cannot measure it, you can not control it, if you cannot control it, you can not manage it”
Objective: better flow of polymer into fine features of mold in injection molding process
Objective: reduce machining (drill, mill, or lathe) costs to make product.
Objective: improve composite part strength
Equipment (or process) variables:
Objective is better control part weight: 25 oz.
Variable 1: extruder temperature
high = 200 C low = 150 C
Variable 2: injection time
high = 4 sec low = 2 sec
E1 – ‘effect’ of variable 1, use the average of “high” minus the average of the “low” values
sleepEffect of Variables?
Average corners of inscribed regular tetrahedrons of diagonals and subtract
One tetrahedron should include the (-,-,-) corner and the other should include the (+,+,+) cornerGraphical Understanding (cont)