Dear experts,
While doing fractional factorial central composite design in Design expert with practical alpha (k>5) for 6 factors, the software default central point replicates is unaltered. How many replicates should one have at factorial and axial (star) points in central composite design? One must perform all the runs right. Your technical inputs and guidance are highly appreciated.
Thanks and regards
Shaileshkumar Baskaran
Normally StatEase recommends that you should have 5~6 replicates of your center point when doing CCD - but it also really depends on how realistic your model accuracy (pred. R²) needs to be for your purpose.
More replicates ~ you fetch more noise = model is more representative
That being said, it won't hurt if you customly add / replicate some factorial or axial points, especially if they're "unexpensive" to run.
Consider if you really need to run a Central Composite Design for 6 factors. A fast check in Design Expert tells me it is about 86 runs, which is tremendous. Adding more blocks into the design increases the number of runs even more.
I don't know how experienced in the design of experiments you are, but here are my suggestions. Firstly, try running 2-level full factorial designs or IV-resolution fractional factorial designs. It will give you the preliminary results and information on which factors are indeed significant for your process. Moreover, you will get an insight into the interactions between factors. Thanks to such screening, you can choose, say, 3 factors, perhaps those which exhibit interactions, for the Central Composite Design. With 3 factors, 6 centre points, 2 replicates for axial and factorial points, divided into 3 blocks, it gives 34 runs. Much more reasonable.
Feel free to PM me so we can discuss it more and maybe I can give you additional tips.
Dear experts,
Thank you so much for the clarifications.'
Regards
Shaileshkumar Baskaran
- Badges
- Science topic
- Similar topics
- Biological Science
- Biota
- Eukaryota
- Fungi