Hi @peddinti_vasu,
Welcome in the Community !
It's very hard to help you without more context, and a possible anonymized dataset to test different models/analysis, as well as evaluating design performances (with Evaluate Designs platform).
Have you tried using the The Fit Two Level Screening Platform ? It usually does a great job at identifying important active effects. If you have created a Plackett-Burman design with JMP, the script "Screening" will lead you to the Fit Two Level Screening analysis platform and enables to explore 2-factors interactions.
Related to your questions, and in the absence of more information, I would recommend to proceed with caution.
Screening and optimization are two different and complementary tasks, with screening being carried out first, often followed by an optimization step.
I highly doubt you can already optimize your system with a classical factorial screening design that only identify important main effects. You might want to dive deeper in the understanding of your system by augmenting your initial screening design thanks to the Augment Designs platform, and specify a more complex model to identify higher order effects, like 2-factors interactions and quadratic effects (2nd order), or even higher order effects.
With so limited context and information, I can only recommend augmenting your design by taking into consideration every factor that seem to be important and significant in the screening stage. When you face a main effect showing p-value close to 0.05, it's best to keep it in next step : it's a better compromise to keep a non-important/non-active effect than to remove it prematurely and miss a potential important effect (or an important effect related to this factor).
Hope this response will help you in the meantime,
Victor GUILLER
L'Oréal Data & Analytics
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)
