275535_en-US

liliconcom

DoE design

Hello everybody,

 

i have two concentration of two chemicals. I can vary both concentrations in a range between 1 - 10%. I would now like to find out how the properties of the product are influenced by the variation of the two parameters and how I can maximize them. I am particularly interested in how different ratios of the two concentrations affect the results.

Does a DoE (classic and effective surface design) make sense here? Are 4 attempts enough for the experiment?

 

Many thanks for your help.

 

John

DOERe: DoE DesignWhat does the total Chemistry / Experiment look like? Do the chemicals interact? What about the remaining 98-80% in case you add the two chemicals to a base? Did you consider mixture design?Re: DoE design

The experiment is a polymerization. The other 98-80% are monomers.

Re: DoE Design

So you can independently vary the concentrations of both of the chemicals? Then you do not need to use a mixture design (factors are mixture components).

 

I would use custom design. The number of runs is determined by (1) the terms in the model and (2) the variance in the response. You need a minimum of 1 run for every parameter that must be estimated. Are you factor ranges wide enough to elicit a large effect? Do you expect the effect of both chemicals to be linear? Do you expect the effect of one chemical to depend on the concentration of the other chemical? It is important to define the model carefully, although with only two factors, you can specify a complex model and it still should not require a large number of runs.

 

The variance of the response is important because the larger the variance, the larger the number of runs to average out the variance and get good estimates.

 

You might also want to use the I-optimal criterion for the design. This will emphasize the importance of estimating (predicting) the response when searching for the optimal concentrations.

 

If you have not seen the specific documentation yet, I recommend selecting Help > JMP Documentation Library > Design of Experiments guide. You should also look up the Prediction Profiler, which is based on the selected, fitted model, to optimize the concentrations.

Re: DoE design

Thanks for the answer.

 

Yes, I can vary the concentrations of the two chemicals as desired. But I have already defined a range in which the concentrations should move.

I'm not sure if the effect of the chemicals will be linear.

The effect of one chemical can affect the effect of the other.

The answer (the result) includes mechanical properties such as tensile strength or modulus of elasticity. But also the rate of polymerization.