Response Surface Methodology - RSM - tutorial

[@ajwaldsouza8812]

Thank you very much for this amazing tutorial. Has given me extreme clarity on the topic. Keep doing more of such tutorials!!

[@Vitarka]

Glad it was helpful!

[@Vitarka]

Hi Ajwal, just give me some suggestions on which point you are facing difficult in understanding and solving so that I can try to solve and prepare a video.

[@anandratanamaurya9975]

thank you so much sir much much much useful knowledge

[@Vitarka]

Welcome

[@Shanthan-g]

hi sir, what if the experimental day and randomized data doesn't match with the input parameters. For suppose in experiment we have 15 and 1.05 but in randomized table that we got from rsm have only 10 and 1 how can we write the response value as there is no experimental value for the randomized generated data

[@Vitarka]

Basically, we need to perform the experiments based on the RSM based design of experiment only.

[@prabhaharansa678]

Very helpful lecture sir. It serves the purpose....👍. Really insightful...!!!!!

[@Vitarka]

Thank you

[@ansa5587]

Super! very nice contribution

[@Vitarka]

Thank you

[@shobhitchaturvedi4866]

Thanks for all this wonderful information. Can you make a video solving a research paper on RSD and also explain Box Beheign Design.

[@Vitarka]

As soon as possible

[@john110503]

Waiting 4 ur video

[@nidhidixit4888]

Thankyou so much sir...please make tutorial on advanced optimization techniques like nsga2, pso, anfis , ann, ga etc.

[@Vitarka]

Welcome. Sure

[@john110503]

nice work

[@Vitarka]

Thank you! Cheers!

[@junaidpir6803]

NICE

[@Vitarka]

Thank you

[@carlmanese1581]

Thank you so much sir. I have 2 factors that i already found the response before using rhe software. Can i replace those columns with my factor values?

[@Vitarka]

Welcome, I am not sure about your point. Please refer Montgomery Design and Analysis of experiments textbook you will get lot of information

[@VenkataSivaRaVadlamani]

Thank you for your invaluable insights. do you know how to change polynomial equation to logarthmic equation?

[@Vitarka]

You can it to Linear Quadratic or 2FI during the analysis. It will be available as a drop down if you use any software for the RSM

[@VenkataSivaRaVadlamani]

​@@Vitarkawhat is 2FI mean ? In addition if response is transform to log and conduct again analysis considering log transform response will the equation be logarithmic this time. I use a minitab. Thanks in advance

[@ajaykushwah6793]

Sir it's a absolutely nice presentation. Sir I have a doubt, I have 4 factors and each having 3 levels, which rsm design I can use in btw Box beigan and CCD ?

[@Vitarka]

I hope three levels (you have to give low and high) where as medium level is automatically taken. I cant say which is best one. It depends on whether your experimentation factor.

[@irfanahmadansari1098]

Thank you for your sincere effort. Can it possible to take two independent variables which are related to each other, like percentage of filler and polymer in a composite.

[@Vitarka]

Yes, absolutely you can use it.

[@shivachannel6812]

Good explanation. Do have any lecture regarding ANN ???

[@Vitarka]

Thank you. As of now no lectures on ANN

[@abhakumari927]

Thank you for the video, sir. If we have been given y in some range(say 78+/-0.2), how do we find the time and temperature such that variance is minimum?

[@Vitarka]

it depends on the minimum and maximum levels we choose.

[@amirsafari7140]

hi, I'm working on laser-assisted turning and currently experimenting with five parameters: depth of cut, feed rate, spindle speed, laser power, and laser angle (the distance between the laser spot and cutting edge). Initially, I opted for a face-centered central composite design. However, I've noticed that when the laser power is set to 0 (indicating the laser is off), the laser angle becomes irrelevant. I'm uncertain about the appropriate experimental design in this situation. Could you please suggest a method or name for this scenario, considering the disregard for laser angle when laser power is zero? Thank you for your insights!

[@Vitarka]

You can add the parameter as 0.1 instead of 0 in the face centered. If 0.1 make sense you can consider in the calculation as negligible. This is thought process.

[@rowys.8414]

Thank you very much sir!

[@Vitarka]

So nice of you

[@inthigenti]

I am trying to analyze an experiment with 3 factors (Mix Ratios) that must tally to 100% on all 9 runs. How do I create the design firstly to tally to 100% and how do I analyze this experiment to know the influence of the 3 factors on the final result after testing such as strength. I’ve been in a dilemma of using the wrong processes and I need clarity on what I should do in this case

[@Vitarka]

Not sure about the Mix ratios. That you are planning for less number of runs (9) Please refer to some textbooks on Experimental design

[@nidhidixit4888]

Respected sir, please reply that if we have 5 levels and 4 factors than how we assign the values of level as low and high...please reply asap.. Thankyou

[@Vitarka]

Hi madam, in the case of 5 levels that is difficult to set in rsm try to use Taguchi method for it

[@vikashkumar-cr7ee]

Hi there, which journal article you considered here for RSM?

[@Vitarka]

Hi it is a generalized example. hope fully you can get it from standard textbook also,

[@drshenbagavelup1345]

Question: If there is any possible to predict signal-to-noise ratio for response surface methodology

[@Vitarka]

Not sure about it. But if we have a Regression equation we can predict in any methodology

[@ansa5587]

have you also a lecture on alias structures ?

[@Vitarka]

No such

[@pranavp1360]

Thank u

[@Vitarka]

Welcome

[@shailavijay1]

Thanks for sharing informative tutorial. I would like to know what's the difference between coded and uncoded value. Do we have any equivalent software package in Python for RSM ?

[@Vitarka]

Welcome and thanks. I hope these packages are not available currently in python.

[@shailavijay1]

@@Vitarka Could you advice which open source Stats tool will give similar results to Minitab for RSM ?

[@Vitarka]

JASP is one such tool, but I havent tried it.

[@kumararaja4330]

Sir how you created the table of temperature and reaction time and X1,X2 is you created after giving values in the software

[@Vitarka]

Yes

[@inthigenti]

How do I select Cube, Axial, and Central points?

[@Vitarka]

It is not explained in the video, you can refer Design of Experiments by montgomery, this textbook will give more insights on how to select the cube axial and central points.

[@ramyadevaraj3168]

How to calculate response values.?. i have only 5 results .. how can i enter these for 13 runs?please answer

[@Vitarka]

Definitely it is impossible to do if you have only 5 results. It is like first you need to design the layout of experiment and then get the results of the experiment

[@m.amareswarireddy7844]

hello sir, thank you for this tutorial. Sir how we can find the optimization parameter

[@Vitarka]

Hi madam, welcome. Here you have to use RSM for Modelling if your parameters. I means you can generate a equation. From that we can perform Optimization or if you have multiple responses you can do gra analysis etc.

[@pkprakash7988]

What is the difference between ccd and central composite rotatable design ccrd

[@Vitarka]

in ccd rotatable design more number of experiments can be performed.

[@nimmiperera8931]

hi,,When I fed the variables for four factors , the combination I got some times deviated to my experimental trials..For that how can I add the response value..Do I want to omit that combinations?

[@Vitarka]

You can mail your problem, my gmail id was vitarkaprojects Basically the response value depends your DoE, later you can type it after the columns or copy paste those.

[@piapsalm7669]

I would to know also,how to compute the response value..where did you get the response(y) value in excel? is it computed?

[@imranwaleed8353]

How do you get the coded values say you had 5 levels what would the -1 and +1 including the alpha values be in the data?

[@Vitarka]

CODED VALUES WILL GET FROM SOFTWARE

[@user-tl1wu6qz9r]

Can you tell how Response(y) is being calculated?

[@Vitarka]

It is is an output response parameter, here I have taken an example which was experimented by some one. Like if you are measuring the temperature as an output we can consider y as output response as example

[@tanujtanwar8417]

Is it possible there could be no quadratic equation means no interaction between three factors

[@Vitarka]

Interaction depends on the coefficient we got. If coefficients are zero then there is no use

[@nimmiperera8931]

hi,, how to recognize how many cube, axial and central points are available

[@Vitarka]

Based on the theory representation. Cube has 8 end points and 6 centre points and one centre point.

[@nimmiperera8931]

Thank u@@Vitarka

[@rohithgowda9437]

Sir, how to identify relative impact of the factors, by comparing the factor coefficient in order to predict optimal input parameters

[@Vitarka]

You mean to say Influence of the factors ?

[@luigi88953]

Hi, my question is about optimization in factorial doe, without recurring to response surface. I was wondering if in order to use the response optimizer in factorial doe, interaction effects are mandatory, or I could also optimize just using main effects. I ask you this because from what I understand in response surface I need at least interaction effects to be able to optimize. I'd appreciate a clarification. Thanks in advance.

[@Vitarka]

It totally depends on the complexity of the problem. Moreover some research problem required only main effects to get optimized. So require interaction effects also. If you observe some of the problems the main independent factors will not have effect but the interaction may effect. Any you question to doe is depend on the number of experiments we are working. This also effect

[@luigi88953]

@@Vitarka thank you. I still have one doubt about this though. In RMS we say that conceptually we need a curvature (interactions and square terms) to be able to optimize. Why can we then still optimize with only main factors in the model, since in this case RMS would have no curvature and a completely flat surface?

[@Vitarka]

Not aware of the curvature of interaction

[@luigi88953]

@@Vitarka interaction terms would twist the surface.

[@luigi88953]

@@Vitarka you are absolutely right to say that interactions don't produce curvature though. What I meant is that they still affect the complete planar surface with just main effects, like they produce a twist.

[@saralahmady1383]

thanks for this rich video, i want to ask you how do i know what kind of design better to use for a study

[@Vitarka]

Depends on the research problem and spend some time on reviewing the literature so that the authors give faults and advantage in applying the design so it will be easy for you.

[@sarasara-yr2xj]

@@Vitarka thanks for your reactivity . plz we need more videos about response surface methodology and how analysis the graphs

[@Vitarka]

Surely i will prepare a videos on it and upload

[@sarasara-yr2xj]

@@Vitarkaplz i have one question why in articles i find the same number of factors and levels but each article has a different number of runs want just know why . exemple in minitab logicial with 2 factors and 2 levels it gives uasuly 13 runs but in articles i find differents number

[@Vitarka]

@@sarasara-yr2xj it depends on number of blocks in the design and number of axial points with that part also the experimental runs will vary

[@divyanshtewari137]

Sir, In my analysis I am getting R-sq around 75.77 and R-sp(adj) as 53.96, What should I do next ?

[@Vitarka]

Sir, it depends on experiment to experiment. If Rsq is 75% it is good fit but not too good. It should be near to 95%

[@piapsalm7669]

hello where I can get the response column? how to compute it?

[@Vitarka]

Hi, the response column in nothing but your output to what ever experiment you conduct

[@satyakrishnacookingchannel6908]

Sir, Thank you for your explaination.I have some doubts regarding the usage of design i.e., when will you use CCD and BOX-Bhenkhen designs, and what are the experimental conditions. kindly give me a reply in this regard.

[@Vitarka]

Sir it all depends on the application level of conduction of experiments. If we need to conduct experiments precisely then prefer BBD.

[@satyakrishnacookingchannel6908]

Sir, Thank you for reply

[@imamadriansyah7962]

Hello everyone, could you please explain me how to find X1 and X2?

[@Vitarka]

Hi X1 & X2 are the levels for your experiment these are coded values depends on the experiment

[@MasudRana-dm6fu]

Nice tutorial on RSM. But, one thing is unclear that is from where or which software did you use to get the coded value and the arrange input values? Please make me clear by giving your valuable answer. Thank you.

[@Vitarka]

Thanks for you kind words. There are many statistics software like Minitab, Design Expert. Etc. I recommend to use Design Expert for RSM based CCD designs

[@supreetkumar7113]

please give me Genetic Algorithm optimization technique example

[@Vitarka]

Just send me refered journal paper to my Gmail - vitarkaprojects

[@tamilhero200]

Sir you make video for response surface methodology in L9 table 🙏

[@Vitarka]

Sure

[@tamilhero200]

@@Vitarka thank you sir 👍

[@neelamrathod8193]

Sir,i have used 6 factors,centre point 3 and 37 runs...is this right?got quadratic with all responses and is significant.

[@Vitarka]

Madam I am not sure about the number of runs. In rsm you can customise your problem

[@neelamrathod8193]

@@Vitarkait's a central composite design...using 6 factors aren't an issue right?

[@Vitarka]

no issue, then it is right. you can customize the centre points also.

[@pranavp1360]

How to find the values..whether it is increasing or decreasing in 3D surface plot can u plz help sir. If possible plz make a video on this . How to find values in 3d surface plots

[@Vitarka]

i did not explain those in this video you can refer any journal paper for it sir.

[@indirakar5095]

thank you so much sir it helps me a lot for my solar project. can I get the ppt?

[@Vitarka]

glad you liked it. I dont have ppt with me.

[@indirakar5095]

@@Vitarka ok okay...I sm doing Masters in NTUST....Thank you Sir it works a lot

[@Vitarka]

Drop me any email to my gmail. vitarkaprojects i will share requirements and supporting files in the ppt

[@deepakkumarsinghresearchsc5609]

sir how to decide which point we should take axial or centre

[@Vitarka]

Sir, It is based on the levels of the which we are going to decide.

[@divyakakara2935]

@@Vitarka levels means, make it clear please

[@Vitarka]

@@divyakakara2935 you are choosing the experimental levels, based on that one the Points are dependent.

[@sachin0721]

Nice Video sir, sir can you make video of Box-Benhken design please?

[@Vitarka]

BBD is also similar will try sir

[@sachin0721]

@@Vitarka Sir please tell me for 4 factors and 3 level what should be block option choose 1 or 3?

[@Vitarka]

1

[@sachin0721]

@@Vitarka thank you sir

[@sachin0721]

Sir can you tell me whare i got all details of Box-Benhken method and all feature explaination

[@sivachemist2020]

thanks, can you please share the Excel sheet

[@Vitarka]

i havent saved with me,

[@monikachand6896]

Hello I m stuck in between RSM.I have one doubt If anyone can help me it will be really helpful. Please rpl

[@Vitarka]

Hi mail your problem, so that we will try to help you Gmail - vitarkaprojects

[@muhammadjunaid5166]

Hi can you please share the PDF version of this presentation , Highly Appreciated !!

[@Vitarka]

there is not in the pdf you can take the screenshots of the video and prepare it

[@stanley6304]

Why the experiment with t = 85 and T = 175 should be repeated 5 times??

[@Vitarka]

It is a block in the design. During the design it has to customise how many repetition

[@CalonMagister]

Very confused -_-

[@Vitarka]

This was a short video, you can refer some textbook for better understanding

[@pranavp1360]

Sir is there any telegram group for this channel for clarifying doubts.if available plz share the link

[@Vitarka]

No such group. You can mail us any time we will try to address your quires

[@pranavp1360]

Thank u brother

[@radhikadeore7762]

A lot of confusion about considering coded or uncoded coefficients!!

[@Vitarka]

Use Uncoded for real values, codes for writing research papers

[@ashokbhupathi4790]

how it is download sir

[@Vitarka]

What you need to download sir.

[@ashokbhupathi4790]

Rsm software sir

[@Vitarka]

You go to Design expert website and download the trial version

[@alishorts33]

Box behnken design pe video banaen

[@Vitarka]

Sure

[@alishorts33]

RSM pe kis kis ka kaam he ?

[@Vitarka]

Optimization

[@alishorts33]

Variables ?

[@Vitarka]

@@alishorts33 you are asking variable optimization

[@shahzebarsal4014]

poor way of delivering the concepts.

[@Vitarka]

okay, Thank you.