DOE-1: Introduction to Design of Experiments
Vložit
- čas přidán 8. 09. 2019
- Dear Friends, this video is created to provide a simple introduction to Design of Experiments (DOE). DOE is a proven statistical tool and is known to be superior to the conventional approach of One-Factor-at-a-Time of OFAT! In OFAT, one tries to optimize the settings of various factors one by one. But this has a major limitation of not getting information about interaction of the factors! DOE helps us to estimate effects of various factors along with their interactions! This is the first in the series of videos that we wish to create and share in near future. Our future videos on DOE will include practical application examples of DOE, analysis and interpretation, optimization using DOE, Fractional Factorial Designs. We will also introduce later to Response Surface Designs.
The video will be useful for all those who want to learn the technique of statistically designed experiments and also to all those who wish to prepare for various exams by American Society for Quality (ASQ). Video is created by Hemant Urdhwareshe who is Fellow of ASQ and is certified by ASQ as Six Sigma Master Black Belt, Certified Reliability Engineer (CRE), Certified Quality Engineer (CQE) and Certified Manager of Quality and Organizational Excellence (CMQ/OE).
Thanks sir im using this DOE I'm completed my research in M.pharm thank you sir
Most welcome! All the best!
great example and effective way to explain DEO , many thanks for the explanation
I am glad you liked it!🎉🙏
A very easy to understand introduction, this is what's actually needed.
Thanks Ankit! Appreciate your feedback!
Beautifully explained, I actually ,got the feeling of what is the significance DOE... Sir, you made the difficult things , look simple.
Thanks and you are welcome! I am glad you found it useful.
I had done almost the same(paper folded helicopter dropping from differentheights and angles)DoE in my classroom of my university in 2006 for my MSc QM studies.
Interesting! Hope the analysis and learning was similar.
Yes
sorry , masters in environmental management ,
who was MSc ?
I never seen such a humble person like you
Thank you so much!
Very helpful! Thanks a lot! I love the pace of the presentation!
Welcome! Appreciate feedback!
Thank you sir, this is really easy to understand explanation. Nice lecture from you.
I am glad you found it useful!
Nicely explained, DOE will be Easier with such well/simply Explained videos...Thank you,Sir.
Keep watching!
Thank you so much for this video! It helped a lot!
I'm so glad!
Sir 100% Valuable and by learn this now i am able to do DOE for spot welding
Thank for your feedback! I am glad you found it useful.
Excellent explanation with example for clear understanding.
Thanks! Appreciate your feedback.
Dear Hemant, thank you so much for this amazing series of videos on DoE. May I ask you a question about this video? How many trials should we do to make sure that errors due to external factors are minimal? i.e. in this video, you dropped the card only once at each Height/Angle combination. However, suppose there was a little bit of breeze in the room or your hand a minor tremor during the drop, it would have affected where the card lands. So to cater to these scenarios, I guess we have to repeat the experiement at least a few times. I am wondering if there is any guidelines or best practices around this? Appreciate your response in this regard. Thank you.
Hello Deepu, This is an introductory video and I did not want to complicate it. One needs to check power of the experiment in case required to decide number of replicates. Power equals probability of detecting an effect when it really exists. (Power=1-beta risk). If the effect is large, less replicates required, if effect is small, more replicates are required. Hope this helps.
Thank you! very helpful and easy to understand!
Thank you!
Very well explained sir. Thank you so much
Thank you! I am glad to know that you found it useful!
Thank you Sir for a clear explanation with example
Glad you liked it!
You are a genious!!!!
Wow! Thank you! Appreciate!
Thank you for this video it will help me explain my assignment in my masteral degree
I am glad you found it helpful! Appreciate your feedback!
Thank you for the video ! If somehow you have a response maximum which is not at one extremity of the factor ? Is it taken into account or do we miss some piece of information with this technique ? For instance, an efficiency increasing with a mass flowrate until a certain point and then a decrease of this efficiency to reach the maximum considered flowrate. If we only consider minimum and maximum flowrates, what happens ? Thanks !
Hello "Whoismarcel", appreciate your interest. Reponse maximum may not be at one extremity of a factor. It it falls in between the low and high settings, the regression equation provieds you optimum settings. Also, many designers add a center point to check for linearity of responses. In case of non linearity, you will need experiments to model nonlinear response(s). This technique is called "Response Surface Method" (RSM). I have not yet made videos on RSM and may make these in future. I hope this answers your question.
With best wishes..Hemant Urdhwareshe
This video is very useful for beginners like me. Thank you.
You are welcome!
Thank you! for this informative lecture on DOE. I have a doubt:
Do the factors need to be independent of each other? For example: if the area and perimeter of geometry are two factors and response has to be measured by varying the area and the perimeter, is it possible to do so? (here area and parameter are dependent but it is also possible to vary area and perimeter independently ie by keeping one-factor constant, other factor can be varied ).
Yes Ridhima! The factors need to be independent in factorial designs.
Very Professional lectures
I am glad you liked it. Appreciate your feedback!
thanks for your explaination.
I am glad to know!
Very informative video!
Thanks! Appreciate your feedback.
Sir, how if i want to make an experiment involving continuous values? I don't think using high and low level is suitable?
You can use high and low levels. In case you suspect noninearity in response, center point can be added.
Awesome explaination
Appreciate!
Thank you!
Hello Sir, you made it very easy to understand. Can you send me entire series of videos repairing Minitab ?
Thank you! There are already few videos on how to use MInitab for Full and Fractional Factorial designs. I will be doing these gradually as and when appropriate, but my preference is to make videos generic and without using special software. Thanks for your feedback.
Nice.
Thanks Mahesh!
The video is really useful
Glad you think so!
You are legend sir
I am overwhelmed! thank you!
The numbers change for the last graph and it's confusing. Why are the effect values 42 and -1 and you plot that? Why did we calculate the differences of the averages for each individual factor and then combination if in the last plot we aren't using those values.
Thanks for your question and keen inetrest! I amtrying to answer as I understand your question. The differences of averages are calculated to analyse 'effects' of the factors and interaction. These values are used to create a 'model' (or sometimes called transfer function) of the process. The model is then used to find optimum settings of the factors. Some of these is explained in the subsequent four videos on DOE.
The coding that you've done is based on what book or logic.? Is there any book that suggests to do it?
Thanks! You can refer to Douglas Montgomery's book. There is a separate video on details of coding. Here is the link:
czcams.com/video/W-H8b86zJcI/video.html
Thank you
You're welcome!
By using different cards each drop are we not adding a new variable each time? So how can we compare the results against each other?
You could drop the same card. By changing the cards, this becomes a noise factor.
@uhemant1 Does it not introduce a new variable? You mention changing the card each time thus introducing a new factor/variable each time. This doesn't make sense to me....
It may not make sense to you. But in every experiment, there are controlled and uncontrolled (noise) variables. As an experimenter, you need to decide which one you may include considering the reality.
Nice explain
Sagar - Great that you’re interested in DOE - It is the fastest way to find out and a key tool for an engineer. Take a look at my video showing you how to plan a DOE correctly…
czcams.com/video/gsD8V2_eZ0A/video.html
Thanks for liking
air flow in room also could be a one variable?
Yes. It can be. But if it is not controlled, then it is a 'noise factor'. If it can be controlled, it can become third factor.
@@instituteofqualityandrelia7902 Thanks for clarification sir,
Sir, could you please explain the conclusion of the experiment? Thank you
The conclusion of the experiment is to understand the effect of the two factors and interaction. This was introduction to Design of Experiment. Based on the effects, we can develop a regression equation and find optimum settings of the factors for desired results. This is explained in another video which you may like to watch. czcams.com/video/W-H8b86zJcI/video.html
How can you prove the response is linear? Could the response not be a quadratic and you missed it?
Hi! Thanks for your question and keen interest! You need to add a center point to validate linearly. You may like to watch related video. Here is the link: czcams.com/video/60WRlk0HEFE/video.html
@@instituteofqualityandrelia7902 thank you
Who are here BE mechanical third year?
Didn't understand your comment.
@@instituteofqualityandrelia7902 can you make more video on design of experiments