Notes

Companies find it hard to produce things with low variability, this is one the most important goals - see toyota model example

inventory the effect is obvious, however in service it could be things like errors made

crit to quality characteristics include: things that are physical, sensory, and time oriented.

differences in products occurs from variability, large variability is no good, while small variability is generally okay. this is direct conq. of differences in the material, variations to the operation process etc.

to quantify quality and variability we use statistical models in order to show that some variables are adherent to specific specifications or not

specifications by manufacturers are bound by an upper and lower limit that determine the acceptable amount of variability that can exist in a given variable.

over-the-wall approach to design includes prototyping into design specifications that are given to the manufacturer. in this regard specifications are set without adherence to the general manufacturing procedures - can produce a lot more nonconforming parts, AKA if serious enough it is considered a defect.

status quo implications warrant use of design team with manufacturing operators to make product - called concurrent engineering.

Shewart created statistically control charts

wartime greatly amped up the need for statistical control

three major specifications, stat process control, design of experiments, and (lesser) acceptance sampling

cycle time is an important CTQ

Statistical process control charts have upper and lower limits to show where a variable should lie in between and the line in the middle is the line that determines how much variability exists in the variable based on the finite actual amount.

CTRL charts is part of process monitoring technique, more you cover it the better you can be at reducing variability

one of the other ways to control variability is the use of designed experiments where you vary the determining effects and see the consequences of what happens, one of the ways is called factorial design where more than one variable are tested together in all possible combinations to see the effects of each combination of factors

We can use the square and cube analysis to show how it'll effect certain ways and to determine which ways are best e.g. book example is back end of the cube is the best way to run the certain operation - reduced variability

designed experiments unlike the process control charts are off line statistical quality tools where the majority of the experimentation occurs outside of the manufacturing process

the routine that allows the in process correction to acceptable values with the use of statistic process control charts can be referred to as engineering, automatic, or feedback control

When it comes to acceptance sampling this is one of the oldest ways to measure qual control and it involves seeing if a product is acceptable in generally two places, the first is in the raw material stage and the second is the final product stage

if a scapped product is reformed and then added to the good amounts its called rectifying inspection

problem with acceptance sampling is that it doesnt tell you shit about the actual process and how it works

remember that the overall goal of quality engineering is to reduce the amount of variability - maximize the reduction - this is key, even if a product is okay we can still make it better

Quality planning stops waste, listens to customer voice - improves on products only that meet and exceed customer standards

quality assurance - quality is maintained, policy, procedures, work instructions, specifications, or records

Q control and improvement is ensuring processes meet requirements and continue to improve - involves SPC and designed experiments.

Deming has 14 points that explain statistical crap, strong emphasis on organizational change

ISO determines is some companies are quality sufficient, problem is they gotta choose someone to conducts the audit so I mean the person could just not care and be like meh, here's your award chump L O L

Six sigma, Motorola late 1980s, focus is to reduce variability in key characteristics - to the point where failures or defects are extremely unlikely

For example, looking at 3 sigma we get that 99.73 percent is inside the specification limits

3 sigma relates to 2700 parts per million being defective
under six sigma quality efforts you get 2 parts per billion that are defective

6 sigma designed that even if a mean shift of 1.5 sigma occurred the overall change would only be up to 3.4 ppm, which is very low

process performance is not predictable unless process behavior is stable

Six sigma improves performance and cost

Six sigma uses 5 step problem solving approach called DMAIC