Hybrid Central Composite Design — Flexible RSM Experimental Design
Hybrid Central Composite Design for Response Surface Methodology · Also known as: Hybrid CCD, HCCD, modified central composite design, hybrid RSM design
Hybrid Central Composite Design (Hybrid CCD) is a class of response surface designs introduced by Roquemore (1976) that combines the structural properties of classical central composite designs with modified or reduced point configurations to achieve rotatability or near-rotatability with fewer experimental runs than a standard CCD, making it especially practical when the number of factors is three to six and experimental resources are limited.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Hybrid CCD when you need to fit a full quadratic response surface model with three to six continuous factors and either the standard axial distance of a spherical CCD produces infeasible factor settings, or you need to reduce total run count below what a standard CCD requires while retaining rotatability. It is well suited to process optimization in chemical engineering, food science, pharmaceutical manufacturing, and materials testing. Do not use Hybrid CCD when you have only two factors (Box-Behnken or standard CCD suffices), when factors are categorical (switch to a split-plot or mixture design), when the response is clearly non-polynomial (a metamodel or machine-learning surrogate may be more appropriate), or when factor ranges are so wide that a quadratic polynomial cannot adequately capture the response surface over the full region.
Strengths & limitations
- Achieves rotatability or near-rotatability — uniform prediction variance in all directions from the design center — with fewer runs than the corresponding standard spherical CCD.
- Keeps all design points within a hypercube, avoiding the extrapolation implied by long axial arms that may be physically infeasible or operationally hazardous.
- Supports estimation of all quadratic, interaction, and linear terms needed for a full second-order model.
- Flexible: multiple catalogued hybrid variants allow the experimenter to choose the best match for their factor count and run budget.
- Center-point replicates provide a pure-error estimate for lack-of-fit testing without inflating the overall run count substantially.
- Roquemore's catalogued hybrid designs cover specific factor counts (typically three to six); designs outside this range require custom construction and specialist software.
- Assumes the true response is reasonably approximated by a second-order polynomial; highly nonlinear responses may require augmented designs or alternative metamodels.
- Less widely implemented in standard statistical software than classical CCD or Box-Behnken designs, requiring familiarity with design tables or specialized packages.
- Sensitive to missing observations: because the hybrid geometry is carefully balanced, losing even one design point can compromise rotatability or reduce prediction efficiency markedly.
Frequently asked
How does Hybrid CCD differ from a standard Central Composite Design?
A standard CCD places axial points at a distance alpha from the center, which for rotatability often exceeds 1 in coded units, pushing those points outside the factor range used in the factorial portion. Hybrid CCD restructures the point geometry — sometimes merging or repositioning axial and factorial points — so that rotatability (or near-rotatability) is achieved without exceeding the hypercube boundary, at a comparable or lower total run count.
When should I prefer Box-Behnken over Hybrid CCD?
Box-Behnken designs are generally simpler to implement, are widely available in standard software, and also avoid corner points (so all points lie within or on the hypersphere). Choose Box-Behnken when you want a well-known, software-supported design with moderate run count. Prefer Hybrid CCD when you need rotatability or have consulted Roquemore's tables and found a variant that fits your specific factor count and run budget better than Box-Behnken.
How many center point replicates should I include?
Three to five center point replicates are typical. They provide a pure-error estimate independent of model lack of fit and improve the estimation of the intercept. The exact number can be chosen to improve orthogonality or to balance the overall design efficiency; consult the specific Roquemore design table for the recommended center-point count for your factor number.
Can Hybrid CCD be used for mixture experiments?
No. Hybrid CCD assumes unconstrained continuous factors that can be set independently. Mixture experiments have a sum constraint (all components sum to 1), which requires dedicated mixture designs such as simplex-lattice or simplex-centroid designs. Using a Hybrid CCD for a mixture problem would violate the sum constraint and produce invalid results.
Is Hybrid CCD available in standard software like JMP or Minitab?
Standard implementations vary. Minitab and JMP focus primarily on classical CCDs and Box-Behnken designs. Hybrid CCD variants may need to be constructed manually from Roquemore's published tables or through specialized RSM packages (e.g., the rsm package in R). Researchers should verify design coordinates before running experiments rather than relying on default software menus.
Sources
- Roquemore, K. G. (1976). Hybrid designs for quadratic response surfaces. Technometrics, 18(4), 419–423. DOI: 10.1080/00401706.1976.10489473 ↗
- Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. M. (2009). Response Surface Methodology: Process and Product Optimization Using Designed Experiments (3rd ed.). Wiley. ISBN: 978-0470174463
How to cite this page
ScholarGate. (2026, June 3). Hybrid Central Composite Design for Response Surface Methodology. ScholarGate. https://scholargate.app/en/experimental-design/hybrid-central-composite-design
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Box-Behnken DesignExperimental design↔ compare
- Central Composite DesignExperimental design↔ compare
- Response Surface MethodologyExperimental design↔ compare