Revisit the “Plan and Run Experiments” Step

Revisit the "Plan and Run Experiments" Step

After you have run your own project(s), you may recognize that your lab data are highly correlated. 

And you may ask, how you could reduce those correlations.

A good method is to use industry proven technology of the Design of Experiments.

With FastR&D software, you have good news, you do not need to formally learn Design of Experiments. if you have already learned that great.

FastR&D has essentially automated the most popular types of design of experiments that you are likely to use for your "routine" R&D lab use. These are typically lab experiments that you can finish in a few hours or a day or so. If your experiments do not fit "this pattern" you may share your needs via a question on our website Chat or ask an expert/consultant that may be available at your Company.

While it is not necessary for you to learn/know the type of designs that FastR&D generates, here is a partial list of the commonly used designs. More design types will likely be added in future versions. Current version handles only continuous variables, next version will include discrete variables also.

  • 2**n Full and fractional factorials
  • 3**n Full and fractional factorials
  • Response surface methods
  • Sequential Assembly and 
  • D-Optimal Designs

With FastR&D software generating a design for your needs is really simple:

  1. 1
    Select a Project Name
  2. 2
    Prepare a Project Description file
  3. 3
    Prepare a Project Definition file
  4. 4
    Start FastR&D software
  5. 5
    Answer two questions.
  6. 6
    Upload two files: (a) Project_description_file and (b) Project_definition_file.
  7. 7
    Go to the DOE tab near the top of the software page.  FastR&D software will attempt to generate a small default "linear" design for three variables, if possible.
  8. 8
    Select the slider for the "Number of Control Variables"  Move it to a desired value.
  9. 9
     Select the slider for the "Number of Experiments"  Move it to a desired number that you would like to run. I suggest that you select a number that is a multiple of 4, if possible. Your basic design is generated and ready for download. 

In a few cases, you may need to generate a "Quadratic" design or a "Full factorial Design." If so you could change the option using the "Choose a design/model" dropdown box. Your design will be updated.

If this is the first time you are generating such designs, or you are not confident enough, ask a key question:

Is the cost of doing so prohibitively expensive or will this design take a very long time? If answer is YES, DO NOT run this design. Consult Tech Support or a local resident expert.

FastR&D recommends: "Fail Fast, Fail Frequently, Fail Frugally, but fruitfully."

Your initial design in most cases should be 8 to 16 experiments with 4 or so additional repeats to establish experimental reproducibility. Ideally at most one to two weeks of experimental effort;  may be even less, if experiments take only a few hours.

Collect the laboratory experimental data.

Prepare the Project_Data file.

And run the FastR&D software. See Next Lesson!

This concludes this lesson.

{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}