without waiting why not use the tools of MATLAB to initially perform the experiments and set the parameters. may be u can hybridize few tools also. latter the best approach can be coded in any of the suitable programming language

Matlab software is very usefull such as GA and ANN. 

Minitab is important software for design of experiment. I suggest you use a combination of these two programs.

I would use RSM or a Mixture design. The optimization is just a simple "simplex" optimization. No need for ANN or GA. 

What factors are you trying to test? 

How are you trying to test them? 

Dear Andrew Ekstrom and other scientists,

In my study, I would like to try 4 factors (MeOH/RPO molar ratio, catalyst content, time and temperature) for transesterificartion. Actually I am using EED software in MS Excel to do. This software is free for using but I think that this software is not perfect, it can not show full of information that I want. I have read some publications. Many researchers/scientists use many expert software, such as Design Expert 7.1, design expert 8.0, Minitab software version 16.2.2 and so on. 

So through this connection, if possible please tell me how I can have above software. I am so happy if you send any software for me.

My email: [email protected]

Thanks for your useful comments so much. Hope receive more helps from all of you

Best regards,

Most of the time, a university will have some software deals. I get Minitab and JMP for free through my university. I am a beta tester for Design Expert too. So, I get that for free as well. 

If your department/university does not have some type of deal, there are some websites where you can rent or purchase software for a discounted price. I have purchased minitab for $99 before. That would be one option. 

Design Expert offers a free trial period. After the trial is over, you can usually ask for a trial extension. The owners of Stat Ease, who makes DX software, have a couple books out. They give you 180 day trials of their software with the purchase of the book.

If that is not enough, you can always use R. There are lots of good references for  DOE work in R. R is free. You just need to learn how to program it. (I use R as a replacement for MatLab.) 

I talked to some people at Stat Ease. They offer reduced prices on their software for students and profs. Go to http://www.statease.com/software/dx9-academic.html and see what types of deals they have.  

Actually you should have posted your query under relevant question.

From your question, I understand that you have chosen the variables, then Full factorial design is an appropriate choice. In Minitab, go to Stat>DOE>Factorial>Create Factorial Design, then input the variables names and their levels. Select the design (for runs), blocking if required and replications you want.

Hey Paul,

What you are looking at would probably be best tested as a Mixture Design. Suppose that you want to test different proportions of sand, gravel and cement. In a batch of product, we can call this 100%. We can have cement be between 50% and 100%, Sand between 0% and 25% and Gravel between 0% and 50%. Then we can test use a constrained mixture design for this.

If you know that A + Cement = 60% of a batch, B + Sand = 30% of a batch and C + gravel is 10% of a batch, then we can create designs for this too. (and fairly easily).

If you know that A + Cement = 50%-70% of a batch, B + Sand = 20%-40% of a batch and C + gravel is 0% to 20% of a batch, then we can create designs for this too. (But it is a bit more difficult. ) 

Do you have any process variables you want to test? Say, mix rate, gravel type, sand type, etc?  

Dear Dang Nguyen Thoai, 

you can use Matlab for this purpose. Matlab has a rich set of toolboxes which allow you to design RSM and ANN-GA.