Intro to Design of Experiments Part 1
I’ll begin posting Tutorials on Design of Experiments.
This is different from the typical design of experiments training.
Intro to Design of Experiments Part 1
Please leave comments on how these tutorials can be of more value to you.
[leadplayer_vid id=”5257FE28D911C”]
Lots more to come.
Subscribe for more.
Here is the transcript:
—————————————
SLIDE ONE: DOE Deep Dive: First, the Basics – QbD Works
Sun Kim, PhD
Presenter: Welcome to DOE Deep Dive, first the basics.
SLIDE TWO: Key Concepts – The Scientific Method model for Experimentation
– Experimental Approach (OFAT, Factorials, others)
– When to use Factorials
Presenter: Today’s key concepts are the scientific method model for experimentation, secondly, experimental approaches, one factor at a time factorials, et cetera, and finally when to use factorials most importantly.
SLIDE THREE: What is Experimentation?
Before we talk about design of experiments specifically, let’s talk about experiments in general. First, you want to ask, why do I want to experiment as a scientist in the industry or in academia? And then, take a few seconds to answer to yourself, how can you define experiment in your own language? Let’s first talk about why. Why, so why experiments? As a scientist I like to predict my process. I like to foresee what the outcome of my process is after tweaking a few knobs, here and there, process parameters, and so forth. The experimentation is a manipulation of controllable factors at different levels to see their effect and responses in the face of noise. So, basically, if you have a process…and we call this a black box or a biologics…you control, these are controllable factors, basically knobs you can turn or buttons you can push, and you can change the levels of these parameters. For example, centrifugal time, nutrient type, and inoculum concentration; and then we’ll see a response which in different measures that we’ll take in, in this case, protein recovery, protein quality. These are called output responses in the face of noise. Of course, there are many other things that you cannot control on a regular basis. For example, the type of operators, the number of operators that will be producing these outputs, as well as, the vendors, they will change over time. You cannot always see what…how the vendor is producing the raw materials, and so forth.
In QbD terminologies, these are called the process parameters (PP), sometime critical, based on how you define critical, so it’s CPP. Outputs basically will be the Y’s in the QA, quality attributes, sometimes depends on how you define critical quality, then it will be critical quality attributes. In Lean Six Sigma world, that’s called Y or CTQs, and, as well as these are X’s, and the noise, and you’ll have some, this will depend on material attributes, which is another terminology use in QbD will be either included in the control factors or the noise factors, but there is always noise, as you’ve seen in the previous case.
So, when you change the levels of all these input factors make sure it makes a bigger response on the Y’s quality attributes than…so that it can overcome the level of noise, so you want the signal to be this. So, the only signal you’ll be seeing is this, probably, and the other is the noise, and so forth. So, there’s a very important concept here.